ASSOCIATE LABORATORY INSTITUT FOR SYSTEMS AND ROBOTICS

LISBON

ISR-LISBONIMAR - Azores

IN+CREMINER

SYNTHESIS REPORT2002-2004

ISR – Torre Norte, IST Av. Rovisco Pais, 1049-001 Lisboa, PORTUGALFax: 351-21-8418291Tel: 351-21-8418289E-mail: jjss@isr.ist.utl.pt

1. Introduction

The Associate Laboratory ISR-Lisbon started his activities in January 2002. The activities were developed in cooperation by the four research units that integrate the ISR-Lisbon.

During his yet short existence ISR-Lisbon has faced a very difficult financial situation. In fact, for nearly one year and a half (from February 2002 to November 2003) the FCT did not transfer any programmatic funds as prescribed in the contract signed with ISR-Lisbon.

During this period the ISR direction (together with the direction of other Associate Laboratories) has met several times with the authorities responsible for the funding, namely the President of FCT and the manager of POSI (Operational Program for the Information Society), in order to solve the problem that was being faced. By demand of FCT and POSI, ISR-Lisbon has produced several budgets, cost statements and reports according to formats that were successively changed. Despite having always satisfied the increasing bureaucratic demands we are still far from having at ISR a normal and stable research atmosphere.

Along this long and painful process we do not seem to have been successful explaining the Ministry of Science and Higher Education that the new bureaucratic methods are not appropriate to deal with scientific institutions. The uncertainty and unpredictability of this whole process of funds transfer, led the institution and, in particular, the young researchers to a state of disbelief that is having very damaging consequences on the usual institutional atmosphere and certainly in our ability to attract new research students and to strive for excellence.

To understand the serious financial situation the ISR-Lisbon was pushed into, we display in Table 1, below, the programmatic funds that have been contracted with FCT and the funds that were in fact transferred with the corresponding dates of transfer.

Year2002 2003 2004 2005 2006

Programmatic funds contracted

with FCT (in euros)*

678 365 882 872.3 1 087 379.42 1 182 151 1 386 658

Programmatic funds

transferred by FCTand

correspondent dates

364 122.47 (Feb.2002)

20 000(Sept. 2003)

+212 770.31(Nov.2003)

Table 1

* According to the contract signed by FCT, ISR-Lisbon and IST.

We recall that the programmatic funds contracted with FCT include the funds allocated to ISR-Lisbon, when it was awarded the Associate Laboratory status, directed specially to hiring new researchers and technicians considered necessary to ensure the fulfillment of the additional activities assumed as an Associate Laboratory.

In particular, the contract involved a scheduled plan for the hiring of researchers and technicians as displayed in the following table:

1

Year

HumanResources

2002 2003 2004 2005 2006 TOTAL

ResearchersWith a Ph.D. degree

1 3 3 4 3 14

Technicians

2 2 4

Table 2

The breach of contract by FCT, regarding the transfer of funds patent in Table 1, led to the impossibility of pursuing the plan for hiring researchers and technicians. Consequently, it inhibited, up to now, the fulfillment of the activities as planned.

As a matter of fact, the President of FCT himself has suggested postponing the hiring of new researchers.

This process severed the institutional trust required for medium and long term commitments regarding hiring human resources and blocked the possibility of ISR-Lisbon contracting potential young researchers of top quality. We look forward to the return of regularity in the institutional honoring of the contracts.

It is also important to emphasize that, in opposition to the previous practice, now the funds are only transferred after expenses have been made and justified with documents (invoices). Since research institutions have not in general extra resources to accommodate this type of procedure, institutions reduce their expenses and consequently the transferred funds reduce accordingly. This process tends to converge to zero.

Despite this abnormal situation, the direction and the thematic areas managers have encouraged young researchers to keep their enthusiasm and energy focused on the research activities. The activities reported ahead are the result of this institutional culture and effort.However renewal of research students is showing now to be more difficult not only because the institutions do not have grants to offer, but also because students are aware of the instability the scientific research institutions are going through.This Synthesis Report provides a summary of the research activity of faculty and research students developed during 2003.

summarizes the activity carried by the Laboratório Associado (Associate Lab) of ISR, during the period from 2002-2004. Its organization reflects the the internal organization of the Associate laboratory in terms of Thematic areas and tries to consider the topics suggested by FCT.Complete details for the years of 2002 and 2003 can be found in the Annual reports duly submitted to the FCT.

2. Some general Comments

In order to give a brief characterization of the activities developed by the ISR-Lisbon during the period from 2002 to the present, we summarize in this section some comments that we hope will help to get an overall view of the strategic guidelines followed by this Associate laboratory.

2

2.1. Quality and Relevance of the Scientific Activity

Publications

In the period reported, ISR – Lisbon researchers have co-authored (see the Annex and the 2002 -2003 Annual Reports):

18 Books 2 Book Chapter 159 Publications in Peer-Reviewed, SCI Journals 192 Publications in other Peer-Reviewed Journals / Proceedings of Peer-Reviewed Conferences 57 Communications and Posters in Peer-Reviewed Conferences 8 Publications in other International Journals

International recognition of the research carried out at ISR-Lisbon Associate Laboratory

Citations

In terms of the number of citations of the papers published in peer-reviewed, SCI Journals: more than 100 (this is not the result of an exhaustive search).

International Awards.

Gift to the Earth Prize awarded by the World Wild Life Fund (WWF) to the Director of IMAR/DOP, Dr. Ricardo Serrão Santos, 2002.

Mineralium Deposita Best Paper Award 2001-2002 to Relvas JMRS, CCG Tassinari, J Munhá, FJAS Barriga, 2001. Multiple sources for ore-forming fluids in the Neves Corvo VHMS deposit of the Iberian Pyrite Belt (Portugal): Strontium, Neodymium and Lead isotope evidence. Mineralium Deposita 36:416-427

Luiz Saldanha Prize awarded by IMAR to Ana Colaço "Solvay Innovation Trophy 2003" for "Management Improvement", through the IN+ project

"Solvay Ideas Challenge", http://www.green-wheel.net/ : Manuel Heitor and Maria Jose Francisco, with Paulo Conde from Solvay Portugal.

Dibner Award 2003 – Society for the History of Technology, SHOT, Best Exhibition, through the IN+ project “Engenho e Obra”, Manuel Heitor, with Organizing Committee of the exhibition “Engenho e Obra”.

Prémio TELEciência 2003 (rank 4) - V Festival Internacional do Filme de Divulgação Científica, “ENGENHO E OBRA: engenharia em Portugal no seculo XX”. IN+: Manuel Heitor, with Organizing Committee of the exhibition “Engenho e Obra”.

IAMOT Award 2004 – Excellence in Research Award on Technology Innovation Management, Intl. Association for the Management of Technology.

INTERNATIONAL TELECOMMUNICATIONS SYMPOSIUM 2002 Best Paper Award to João Xavier, Victor Barroso, and Paulo Lopes. Blind Source Separation and Channel Identification: Exploiting 2nd-Order Statistics in Bayesian Frameworks, PROC., ITS2002

Invited Plenary talks

3

During the period from 2002-2004, the researchers of the Laboratório Associado were quite active in delivering plenary talks in Conferences and Symposia as well as invited lectures in Workshops and other international meetings, as detailed in the Thematic Areas reports and in the annex. Overall, the number of events in which the researchers participated can be broken down by years as follows: 11 (2002), 26 (2003), 18 (2004).

Current importance of the research themes; involvement in new, emerging areas of research

The research topics under development in the Associate Laboratory reflect the expertise of the partner institutions and are organized in terms of Thematic Areas:

Thematic Area A-Technologies for Ocean Exploration (ISR-Lisbon, IMAR Azores, CREMINER)

In this thematic area it is central the R&D activitiues the development of methodologies and equipment that will completely change traditional methods of ocean exploration. This trend is explicitly acknowledged in the Integrated Marine Science Plan for Europe that was given priority consideration under the 6th Framework Programme and is being instrumental in defining guidelines for the 7th Framework Programme.

Thematic Area B-Monitoring and Surveillance using Robotic Agents (ISR_Lisbon,)

In this thematic area the central topics are Modelling of biological systems, research on cognitive systems (involving neuroscientists), emotion-based robotics, hybrid systems, spacecraft formation guidance, navigation and control for planetary observation by multi-aperture interferometry-based telescopes.

Thematic Area C-Sustainable Technologies and Environmental Systems (IN+)

In this thematic area the central topics are Energy Systems; Environmental Systems; Technology Policy and Management of Technology.

Further involvement in new areas has been limited by the absence of the planned funding through the Associate Laboratory, namely in terms of post-doctoral researchers

Thematic Area D-Sustainable Technologies and Environmental Systems (ISR-Lisbon)

Signal, image and video processing techniques find applications in several areas of practical interest. The large technological development of the digital signal processing electronic systems has enabled the increase of their processing speed and of their integration capacity as well as the decrease of the costs. Since this tendency is expected to be even clearer in the future, the signal processing techniques appear as one of the fundamental tools to the development of the future communication systems.

Multidisciplinarity and relevance for other research areas and/or technologies

The activities of the Associate laboratory are multidisciplinary, linking basic and applied research to technology development, and focused on the issues of sustainability, namely in terms of the needs to secure the quality of the environment, together with the management of energy resources and the economic development.

Internally and just to give an example both the thematic areas A and B plan to work with cooperating robots that constitute a robot network, which requires advanced communication techniques based on efficient signal processing methodologies, topic of research of Area D.

Other examples can be found in the Thematic areas annex.

Involvement of young researchers and/or post-docs in the research activity

The involvement of Post-doctoral researchers was severely limited by the absence of the planned funding.

4

However more than 20 Ph.D. students have concluded their degrees and a large number are pursuing their Ph.D. programs.

Organization of scientific meetings and advanced training seminars

The transfer and diffusion of knowledge have been particularly promoted through the organization of international Conferences, which have considerably contributed to diffuse knowledge worldwide, and promote the internationalization of the Portuguese S&T system and the organization of advanced seminars:

During the period from 2002-2004 the Associate Laboratory was responsible for the organization of 20 scientific international meetings.The Lab was also involved in the co-organization of other national and international meetings.

International Conferences

2002

o International Symposia on Applications of Laser Techniques to Fluid Mechanics : http://in3.dem.ist.utl.pt/conflaser/

o Ocean Drilling Program Leg 193 Second Post Cruise Meeting. The Ocean Drilling Program Leg 193 Second Post Cruise Meeting took place from 14-19 October 2002, in Lisbon, Castro Verde and Aljustrel.

o ICES – (International Council for Exploration of the Sea) - “ Working Group on the Biology and Assessment of Deep-sea Fisheries Resources.” – Museu do Mar (Fábrica da Baleia), Horta, 4-10 de Abril. Azores.

o BIOMARE Steering Committee and Regional Arctic / Atlantic Working Group Meeting. Horta, 17-21 April. Co-Organizers: Ricardo Serrão Santos, Frederico Cardigos, and Fernando Tempera .

o II MoMAR Workshop - Towards planning of sea floor observatory programs for the MAR Region. Horta, 15 - 17 June. Co-organizers: Ricardo Serrão Santos & Ana Colaço (IMAR-DOP/UAç), Javier Escartin (CNRS, Paris) & A. Adamczewska (InterRidge , Tokyo).

o Planning the Management of Deep-sea Hydrothermal Vent Fields MPAs in the Azores Triple Junction. Horta, 18 - 20 June. Co-organizers: Ricardo Serrão Santos & Ana Colaço (IMAR-DOP/UAç) & Sabine Christiensen (WWF-International).

o 10th IEEE Mediterranean Conference on Control and Automation, MED2002”, Lisbon, Portugal, July 9-12, 2002.

2003

o International Conferences on Technology Policy and Innovation : http://in3.dem.ist.utl.pt/confpolicy/

o Marine Sciences and Technologies: The Luso-Indian Cooperation - With participants from India and Portugal, August 13, Centro do Mar, Horta, Faial , Açores. Co-Organizers: Ricardo Serrão Santos (IMAR-DOP/UAç), António Pascoal (ISR-Lisboa) and Elgar Desa (NIO, Goa , India ).

o Long-term monitoring of deep-ocean hydrothermal ecosystems", 15 to 17 October, Barcelona , Exploratory Workshop funded by European Science Foundation (EW02-17). Co-organizers Ricardo Serrão Santos (IMAR-DOP/UAç) and Javier Escartin (CNRS-Paris).

o OGAMP (Planning and Management of Marine Protected Areas). International Scientific-Technological Workshop. Coordinated by Ricardo Serrão Santos & Rogério Ferraz (IMAR-DOP/UAç). 29 -31 October.

o Omnivis03 – International Workshop on Omnidirectional Vision and Camera Networks.

5

o Robotica2003- Portuguese Robotics Open

2004

o International Symposia on Applications of Laser Techniques to Fluid Mechanics : http://in3.dem.ist.utl.pt/conflaser/

o Stakeholders Workshop on Oceanic Seamounts (OASIS: Oceanic Seamounts: an Integrated Study). 1 & 2 of April 2004, Horta, Azores . Co-organizers: Ricardo Serrão Santos (IMAR-DOP/UAç) and Sabine Christiansen (WWF-International).

o Open conference on the Census of Marine Life (CoML) initiative and Ocean Research in Portugal , 2nd of July, Centro do Mar, Horta. Mário Ruivo (Chair), Tiago Pitta e Cunha – National Strategy for the Oceans, Frederick Graslé – The CoML Initiative, Ricardo Serrão Santos – Azores : Where knowledge of the sea deepens!

o 5th IFAC Symposium on Intelligent Autonomous Vehicles, IAV 2004, Lisbon, Portugal, July 5-7, 2004.

o V IEEE SIGNAL PROCESSING WORKSHOP ON SIGNAL PROCESSING ADVANCES IN WIRELESS

COMMUNICATIONS, SPAWC’04, Lisbon, JUL 2004

o RoboCup2004

o PETS04 – ECCV Workshop on Performance analysis and Tracking in computer Vision.

Advanced Training Seminars

DESIGN OF ROBUST MULTIVARIABLE FEEDBACK CONTROL SYSTEMS, a one semester doctoral level course taught by Michael Athans and Carlos Silvestre at the Instituto Superior Técnico, Lisbon, Portugal, 2002.

INTRODUCTION to the SYNTHESIS of FEEDBACK CONTROL SYSTEMS USING LINEAR MATRIX INEQUALITIES, a three weeks open course taught by Carlos Silvestre at the at the Instituto Superior Técnico, Lisbon, Portugal, 2002.

NONLINEAR CONTROL SYSTEMS, a one semester doctoral level course taught by António Pascoal at the Instituto Superior Técnico, Lisbon, Portugal, 2002

DYNAMIC STOCHASTIC ESTIMATION, PREDICTION AND SMOOTHING, a one semester doctoral level course taught by Michael Athans and Paulo Oliveira at the Instituto Superior Técnico, Lisbon, Portugal, 2003.

DESIGN OF ROBUST MULTIVARIABLE FEEDBACK CONTROL SYSTEMS, a one semester doctoral level course taught by Michael Athans and Carlos Silvestre at the Instituto Superior Técnico, Lisbon, Portugal, 2003.

INTRODUCTION to the SYNTHESIS of FEEDBACK CONTROL SYSTEMS USING LINEAR MATRIX INEQUALITIES, a three weeks open course taught by Carlos Silvestre at the Instituto Superior Técnico, Lisbon, Portugal, 2003.

CONTROL OF AUTONOMOUS MARINE VEHICLES, a 6 hour TUTORIAL by António Pascoal, given at the 6th IFAC Conference on Manoeuvring and Control of Marine Craft, University of Girona, Spain, September 16, 2003.

NAVIGATION, GUIDANCE, AND CONTROL OF AUTONOMOUS MARINE VEHICLES, a 4-hour SEMINAR by António Pascoal given at the Department of Engineering Cybernetics, Norwegian University of Science and Technology (NTNU), Trondheim, Norway, September 23-24, 2003.

José M. F. Moura. SENSOR NETWORKS: A GRAPH BASED APPROACH., Departments of Electrical and Computer and BioMedical Engineering, Carnegie Mellon University, USA

Ali H. Sayed. SIGNAL PROCESSING AND COMMUNICATIONS CHALLENGES IN WIRELESS LOCATION.Electrical Engineering Department, University of California – LA, USA

6

Babak Hassibi. DISTRIBUTED SPACE-TIME PROCESSING IN WIRELESS NETWORKS. California Institute of Technology, USA.

Amos Lapidoth. THE TRUTH ABOUT CAPACITY CALCULATIONS FOR FADING CHANNELS. Swiss Federal Institute of Technology, Switzerland.

Luc Vandendorpe. SOFT INFORMATION AIDED PARAMETER ESTIMATION. Communications and Remote Sensing Lab., Université Catholique de Louvain, Belgium.

Jonathan H. Manton. THEORY OF OPTIMIZATION ON MANIFOLDS. Electrical and Electronic Engineering Department, University of Melbourne, Australia.

Anuj Srivastava. BAYESIAN ESTIMATION AND TRACKING OF DYNAMIC SIGNAL SUBSPACES. Department of Statistics, Florida State University, USA.

Jean Claude Belfiore. CONSTRUCTIVE CODING ON THE GRASSMANN MANIFOLD: APPLICATION TO NON COHERENT SPACE-TIME COMMUNICATION Dépt. Communications & Electronique, Ecole Nationale Supérieure des Télécommunications, France

Victor Barroso. BLIND SOURCE SEPARATION IN CELLULAR COMMUNICATION SYSTEMS., IST/ISR, Portugal

Mos Kaveh. MODELING AND EXPLOITING SPACE-TIME CHANNELS FOR WIRELESS COMMUNICATIONS. Dept. of Electrical and Computer Engineering, University of Minnesota, USA.

Takeo Kanade. RECENT ADVANCES IN COMPUTER VISION. Robotics Institute, Carnegie Mellon Unversity, USA.

Sebastien Bausson. DRIFTING SENSORS IN CLASSIC OCEAN ACOUSTIC TOMOGRAPHY. Laboratoire des Images et des Signaux, St Martin d’Heres, France.

Xiang-Gen Xia. ORTHOGONAL SPACE-TIME BLOCK CODES, COMPOSITIONS OF QUADRATIC FORMS, AND ORTHOGONAL DESIGNS. Dept. of Electrical and Computer Engineering, University of Delaware, USA.

Milica Stojanovic. HIGH SPEED WIRELESS UNDERWATER COMMUNICATIONS. Massachussetts Institute of Technology, USA.

John H. Cozzens. AN OVERVIEW OF NSF AND OUR MISSION. NSF Program Director, Signal Processing Systems, Division of Computer-Communication, USA.

João Xavier. LOWER BOUND ON INTRINSIC VARIANCE OF ESTIMATORS IN RIEMANNIAN MANIFOLDS.IST/ISR, Portugal.

João P. Costeira. THE CORRESPONDENCE PROBLEM IN COMPUTER VISION. IST/ISR, Portugal.

Pierre Borgnat. SCALE INVARIANCE FOR SIGNALS AND IMAGES, AND STATIONARIZATION OF PROCESSES. INRIA, France.

João Gomes. TIME-REVERSED UNDERWATER COMMUNICATIONS. IST/ISR, Portugal.

Paulo Gonçalves. WAVELETS, A VERSATILE TOOL: PDES APPLIED TO TIME FREQUENCY DISTRIBUTIONS, AND EXISTENCE TEST OF MOMENTS. ISR, Portugal.

In addition, a major series of research workshops is organized at IST on “Science, technology and Society”, http://in3.dem.ist.utl.pt/adv/workshops/

2.2. Organization and Scientific Environment

Culture of creativity and opportunity/encouragement of younger researchers initiative

Young researchers are strongly encouraged to developed research activities and several post-graduate programmes and courses (in English) are being offered by ISR-Lisbon and partner institutions.

7

Also, undergraduate students are involved at a very early stage of their education in research activities.

Share of common equipment, facilities and resources between different groups Ever since its creation, the different partners and groups of the same partner involved in ISR-Lisbon have shared equipment, facilities, and resources. This is patent in a number of activities that include the following:

Utilization of the Arquipélago and Águas Vivas research vessels (together with supporting speed boats) by the Marine Robotics group of ISR and the group of Mineral Resources of CREMINER during tests at sea in the Azores. The ships are maintained and operated by the DOP/UAzores, who is also responsible for the coordination of the ship crews.

Sharing of Lab facilities in the Azores: During the Summer, the Marine Robotics group of ISR routinely transfers a large part of its Lab to the Azores. The equipment is transported by ship in a container and installed in a warehouse of the IMAR-DOP/UAzores, which provides the basic facilities required for equipment testing, marine robot integration, and transportation to the harbor prior to vehicle and equipment deployment. This support has proved instrumental in allowing the Marine Robotics group to enhance its operational capabilities. The Lab facilities are also routinely used by CREMINER.

Conversely, the IMAR-DOP/UAzores has used the resources of ISR and CREMINER. During tests at sea with the Marine Robotics group, scientific equipment is installed on board the INFANTE AUV and the DELFIM ASC to acquire marine data. The IMAR-DOP/UAzores has also used the X-Ray Diffractometer that is property of CREMINER. Both CREMINER and IMAR-DOP/UAzores use the facilities of ISR during meetings for projects preparation and planning of tests at sea. Facilities are shared by projects involving various teams. Examples can be among others the Rescue project (mobile robotics, vision, intelligent control) and CAVIAR (cognitive vision systems for surveillance)

Over the past years, the groups of ISR located in Lisbon and in the Algarve have shared acoustic equipment for acoustic data acquisition and processing.

Level of complementary activities and research interaction;

Several PhD and MSc students are co-supervised by researchers of diffferent groups. There are projects involving members of different groups. A large number of projects involve researchers of the partner institutions and other international groups.

Organization of seminars and group meetings; level of involvement/participation in those events

At ISR-Lisbon Seminar meetings are organized on a weekly basis. The seminars afford young researchers the possibility of presenting the results of their work. Some of the meetings are more focused on the discussion of open themes for research. Bi-weekly seminars are also organized at CREMINER. Internal Weekly Lectures are routinely organized at the IMAR/DOP. See the Annex.

Internal communication (instruments and level of participation)

INTRADOP – an internal Webpage was set-up for sharing information on marine science and complementary activities throughout the IMAR/DOP research unit (more than 200,000 visits in 6 years).

2.3. Internationalisation

Joint publications with foreign researchers

The list of publications in the Annexes and in the Annual Reports show clearly that the main thrust of the research work carried out in the Lab is done by exploring strong cooperation links with foreign institutions. Also the Associate Laboratory is active in promoting the visit of international researchers.

8

Participation in scientific and technological projects with foreign researchers

During the period from 2002-2004, the Lab was involved in a large number of projects with strong international participation. These projects have proved instrumental in the definition of challenging scientific and technological objectives and have steadily paved the way for the establishment of solid cooperation links with highly reputed institutions in Europe and around the world. See the Annex and the Annual Reports for details.

Involvement in international networks for higher education and research

The institutions that integrate the Lab are deeply involved in research and teaching. This has led naturally to the establishment of cooperation links with international institutions in the form of networks for higher education and research. See the Annexes and the Annual Reports. The networks set the ground for the education of young scientists who benefit from intensive discussions of theoretical concepts and practical solutions to challenging problems in a vibrant atmosphere.

A few examples of major international networks that have been actively participated and/or coordinated during 2002-2004:

Globelics, www.globelics.org :Aim: to improve understanding of the role of innovation and competence building systems for economic development.

Graduate Consortium in “Technology, Management and Policy”

Aim: To foster the academic analysis of Technology, Policy and Management in a way to bring together graduate students and researchers in Systems Engineering, Policy Analysis and Management and major faculties on Technology, Policy and Management.

Cluster Entrepreneuship Task Force:

Aim: to foster entrepreneurial education across Europe

Human Capital Mobility Network – CAMERA under the 5th Framework Programme.

EURON II – European Robotics Network – under the 6th Framework Programme

Interaction with foreign researchers and/or research units abroad

The level of interaction with foreign researchers and/or research units outside the country is visible in the number and scope of the international projects and networks, as well as in the number of joint publications and foreign researchers integrating the Associate Laboratory groups. See the Annex and the Annual Reports.

Ability to attract foreign post-docs and other researchers (invited or associated)

During the period from 2002-2004, has maintained the contracts with two senior foreign researchers (from MIT and the University of S. Paulo) and has successfully recruited several post-docs and Ph.D. students from outside Portugal. This was done in spite of the tight financial situation of the Lab. Its regularization is a key enabling step that will permit, in the long run, the recruiting of a much larger number of foreign researchers.

2.4. Human Resources and Management

Recruitment policy of the “Laboratório Associado”

The recruitment is essentially done by International advertising

9

Programmes for post-graduate students and advanced training of younger researchers

Several Post-graduate programmes and courses are offered (in english) by the Associate laboratory researchers:

o Master in “Engineering and Management of Technology”, http://in3.dem.ist.utl.pt/master/

o Master in “Engineering Design”, http://in3.dem.ist.utl.pt/mscdesign/

o Signal processing in Clinical Neurophysiology, Pos-grad course in Medicine, FML “Biomedical Engineering - BioMed 02” by Agostinho Rosa

o Nonlinear Control Systems - a one semester doctoral level course, 2002, by António Pascoal

o Introduction to the Synthesis of Feedback Control Systems Using Linear Matrix Inequalities – a three weeks open course , 2002, by Carlos Silvestre

o Nonlinear Programming – Doctoral Program course- 1st Semester 2002, by João Sentieiro o Modelling and Control of Automated Manufacturing Systems - Masters Program course- 1st

Semester 2002, by Luis Custódio

o Artificial Intelligence - Masters Program course- 1st Semester 2002, by Luis Custódio o Dynamic Stochastic Estimation, Filtering, Prediction and Smoothing – Doctoral, Program course -

1st Semester 2002 by Michael Athans/Paulo Oliveira

o Design of Robust Multivariable Feedback Control Systems - Doctoral Programcourse – 2nd Semester 2002 by Michael Athans/Carlos Silvestre

o Discrete Event Dynamic Systems - Doctoral Program course- 1st Semester 2002, by Pedro Lima

o Stochastic Processes, PhD Course 36h, by Victor Barroso.

o Non-linear Signal Processing, PhD Course 36h, by João Xavier.

o Statistical and Computational Models of Vision, PhD Course 36h, by Pedro Aguiar (with José S. Vitor).

2.5. Technology Transfer / Diffusion and Valorisation of Results

Involvement on the launching of “spin-off” enterprises

In 2002, a new company called Blue Edge was created as a spin-off of the activity carried out at Associate Lab – ISR in the area of marine robotics, with direct involvement of senior members of its staff. The main purpose of Blue Edge is to enable the transfer of technology to the industry in a number of areas that include: i) automatic data acquisition and processing, ii) sensor integration, iii) navigation and control of autonomous platforms, iv) real-time computer networks, and v) advanced marine systems for scientific and commercial applications. At the same time, Blue Edge serves the role of providing a natural “stepping stone” for young, bright researchers coming out of academia, in their quest to bring their knowledge to bear on the development of innovative products. This role can hardly be overemphasized since this type of researchers is usually looked upon as being overqualified for the types of jobs that are commonly offered in the industry.

Over the past few years, ISR/IST has been involved in a number of projects for marine vehicle development that include partners from the industry. Privileged links have been established with the company of naval engineers RINAVE and with CONAFI, a company specialized in the construction of fiber-glass hulls. RINAVE played a very active role in the mechanical design of the INFANTE

10

Autonomous Underwater Vehicle and the CARAVELA Autonomous Surface Craft. Presently, it shares with ISR/IST and the National Institute of Oceanography in Goa, India, the task of designing the new MAYA AUV. CONAFI manufactured the hull of CARAVELA. Both companies are interested in the commercialization of these types of vehicles.

In previous years ISR-Lisbon was involved in the creation of other 3 companies:

IdMind – Educational robotics and robot integration (http://www.idmind.pt)Observit – computer vision and video surveillance systems (http://www.observit.com.pt/)Reverse – computer vision and 3D reconstruction

Also researchers of IN+ have been involved in fostering C&T commercialization activities in a way to promote the creation and diffusion of knowledge beyond academia, by establishing conditions that will: o Stimulate university entrepreneurship, through student and staff involvement in technology commercialisation projects; o Foster advanced training and qualifications in technological platforms, by combining technical skills with a broader vision of the relationship between new technologies with economy and society; o Promote entrepreneurial projects and the diffusion of applications/contents for new technologies.

Innovative technological prototypes and patents

“A Controlled Thruster-Driven Profiler for Coastal Waters”, a patent awarded by the Council of Scientific and Industrial Research, Rafi Marg, New Delhi, India to Elgar Stephen Desa, Prakash Mehra, Gajanan Purushottam Naik, Bailon Antonio Ehrlich Desa, Rajachandran Madhan, all of the Institute of Oceanography, Dona Paula, Goa, India, all Indian citizens, and António Manuel dos Santos Pascoal, of the Institute for Systems and Robotics, Av. Rovisco Pais, Lisboa, Codex, Portugal, Portuguese citizen. World patent pending.

Research contracts with enterprises

POPA – Observers Program of the Fisheries of the Azores – contracting parts: AICPA (Industrial Association Fish Processors), APASA (Azorean Fisheries Association of Tuna and Similar Fish), LotAçor (Azorean Fish Auction), launched in 1998.

Research contracts with enterprises include those related with auto-parts components and environmental management strategies.

Services contracted and provided to external entities, public and private

Portuguese Ministry of Economy, IAPMEI

Portuguese Ministry of C&T; through the National Observatory of C&T

Portuguese Ministry of Environment, Instituto de Residuos

European Union - Project review/evaluatrion

ESA - Delegate to the Aurora Program for the Exploration of the solar System

GEMAS - Assessment and management of sand resources of the Azores. Environment Secretariat of the Azores (2001-2004).

RIVA - Management of inshore resources of the Azores.Fisheries Directorate of the Azores (1998- ).

MICA - Assessment of Commercially Interesting Invertebrate Species of the Azores. Fisheries Directorate of the Azores (1998- )

BARCA - Artisanal Fisheries of the Azores. Fisheries Directorate of the Azores (2000-2002)

CRUSTAÇO - Experimental fishing of deep-sea crustaceans of the Azores. Fisheries Directorate of the Azores (1997-).

11

SEA TURTLES/ LONGLINE EXPERIMENT - Decrease of Lethal Impact of Long-line Fisheries on Sea-Turtles. Governement of the United States of America and the Environment Secretariat of the Azores (2001-2005).

MINILINE - Introduction of new Fishing Methods in the Azores. Fisheries Directorate of the Azores (2001-2002).

MAREDA - Yearly assessment of relative abundances of demersal species in the Azores. Fisheries Directorate of the Azores (1997-).

Organization of activities aiming for diffusion of scientific culture (schools; public in general)

Science & Technology culture has been promoted through a series of major initiatives for children, youngsters and the population at large. ISR-Lisbon and its partners have been very active organizing events and activities aiming to increase the public awareness regarding the importancev of Science and Technology. The following are examples of such initiatives:

IN+ “Engenho e Obra: Engenharia em Portugal no sec XX”http://www.engenharia.com.pt/Exhibition: 9 January – 2 March, 2003

The film: “Engenho e Obra – Engineering in Portugal in the 20th Century”

“PENSAR e FAZER engenharia com os mais novos”http://green-wheel.innovagency.com/site/gwb_competicao_01.asp?idioma=0&competicaoid=6National competition for basic and secondary education, 2003.

ISR-Lisbon Participation in the Exhibit “ENGENHO E OBRA”, Jan. 8 – March 2, 2003. During the exhibit, full scale models of the INFANTE AUV and the DELFIMx ASC were on display, together with a video illustrating the development and operation of autonomous marine vehicles at sea. The activities carried out at the ISR were further disseminated through a video that was prepared by the exhibit organization and shown on a national TV Channel. Robotica 2003 – Portuguese Robotics Open, held in Lisbon (about 400 participants form 90 teams from all over Portugal, and 75%+ participants from basic and secondary schools)

RoboCup2004 – held in Lisbon., the largest robotics event worldwide (1500+participants, 400+robots, 350+ teams)

“Ciência Viva” activities together with high school students. Scientific occupation of young students during the summer holidays.

IMAR-Azores

IMAR-DOP/UAzores is strongly involved in out-reaching and educational activities. Over the past few years, the Departament pursued the following activities:

Establishment and exploration of an Interpretation Centre on Marine Protected Areas of the Azores. The IMAR-DOP/UAzores is now preparing an up-grade of that centre that will be dedicated to the Habitats and Biodiversity of the Azores Triple Junction / Mid-Atlantic Ridge.

12

Editing of a Newsletter entitle “Mar Salgado” with 1000 copies, which will increase to 2000 due to high demand. Collaboration with the two monthly Portuguese magazines “Mundo Submerso” (dedicated to submarine activities - 12000 copies per month) and “Saber/Açores” (regional magazine focused in the Azores - 15000 copies per month), on the writing of articles on marine life and marine research issues for the public at large. Occasionally, the IMAR-DOP/UAzores collaborates with other magazines such as “Super-Interessante”, “National Geographic-Portugal”, “Visão”, and “Fórum Ambiente”.

Collaboration with the national and regional public television stations in the programs “Ilhas Vivas” (1989: a six episode series on marine life of the Azores), “Maré-Viva” (6 episodes), “Bombordo” (4 episodes), Planeta Azul” (3 programs: 90’s), and “Mar-à-Vista” (15 episodes: 2003-2004). The latter received scientific advice of and was also co-produced by IMAR/DOP/UAzores.

IMAR-DOP/UAzores has promoted 25 initiatives all year round in high schools, professional schools, local administration bodies (Câmaras Municipais, Juntas de Freguesia), fisheries associations, tourist enterprises and other stakeholders. Almost 2000 persons were directly involved in these actions.

The Department has also edited over the years information brochures and didactic material in many distinct formats (leaflets, brochures, T-shirts, posters, flyers, etc), including a Didactic Kit to schools (2002), a DVD on marine conservation (2004) for the general public, and users manuals (2003).

IMAR-DOP/UAzores has two people permanently allocated to outreach and education activities. At the same time it manages the ImagDOP, a section of image and multimedia, and several web sites: www.horta.uac.pt, www.intradop.info , www.popa.info , and www.macmar.info

2.6. Strategic Planning and Long Term Perspectives

See Activities Plan .

13

2.7. Resources for the Research activity

Adequacy of resources for the research activity Adequate Not Adequate

Facilities xLibrary journals x books xEquipment XTechnical support xAdministrative support xFunding X

14

3. Brief description of the Activities developed within each thematic area

3.1.Thematic Area A: Technologies for Ocean Exploration

For the sake of completeness, an Appendix is included that contains new information related to 2004 and, in some cases, complements the info previously submitted for the years 2002 and 2003.

Introduction

Under Thematic Area A, three institutions were brought together to the core of a research and development program that aims to push the frontiers of knowledge in marine science and its enabling engineering fields. Underlying this unifying theme is the need to bridge the gap between marine science and technology. This is being done by exploring fruitful collaboration links among engineers and marine scientists, namely biologists, geologists, and oceanographers. This symbiosis is instrumental in providing engineers with complex, challenging problems in the field of marine technologies. Conversely, it will afford marine scientists ever increasingly complex technologies to explore the ocean frontier, especially in hazardous conditions.

The document that follows summarizes the first steps taken towards cementing fruitful cooperation links among the different players involved in a common endeavor. Part of the activities reported are the natural expression of basic research that is steadily being done on a vast spectrum of areas that range from marine biology, marine geology, and oceanography, to marine robotics, signal processing, and ocean acoustic tomography. These activities are of course complemented by more practically-oriented work that involves the development of new methodologies and equipment for marine data acquisition and remote sensing, the unfolding of experiments and missions at sea, and the processing and dissemination of information to the scientific communities and the population at large. Both types of activities are being carried out in the scope of national and international projects that are the main driving force behind the establishment of strong synergies with highly reputed institutions worldwide.

The proposed research effort is well rooted in previous collaborative links established among the Institute for System and Robotics (ISR), IMAR-DOP/UAzores, and Creminer / FCUL). The ISR contributes to this initiative with its proven expertise in marine robotics, information processing and computers, and underwater acoustic signal processing. Namely, design and development of marine robots, study and implementation of advanced algorithms for acoustic signal processing, development of vision systems for environment reconstruction and classification, and operation of autonomous platforms at sea. The Creminer and IMAR-DOP/UAzores provide the scientific counterparts with their interests and/or expertises in the areas of submarine hydrothermal activity, submarine methane production, accumulation, and dispersion, deep biosphere, marine volcanology, marine biology, and oceanography.

As will become clear, great part of the work is driven by the need to steadily develop the technological means that are required for ocean data acquisition and analysis at a scale heretofore unknown in Portugal. Illustrative examples include the improvement of laboratory facilities to increase analytical capabilities, enhancing the operational capabilities of the Research Vessel Arquipélago, development of a portable laboratory for geological and chemical studies, expanding the capabilities of existing marine robots, development and installation of a wide area network of acoustics and oceanographic sensors for long-term ocean monitoring, deployment of long and short-term acoustic observation systems for ocean acoustic tomography, and development of advanced methods for underwater target tracking, including fishes as well as marine mammals and robots. To a large extent, the program undertaken targets the Azores region (with its complex network of abyssal plains, seamounts, hydrothermal vents, island margins, islands, and a very large extension of the Mid-Atlantic Ridge) as a natural laboratory for the study of a number of challenging scientific issues. In fact, the study of phenomena that occur in the area holds the key to a better understanding of the interactions between physical, biological, geological, and chemical processes in the ocean.

Again, central to these studies is the development of methodologies and equipment that will completely change traditional methods of ocean exploration. This trend is explicitly acknowledged in the Integrated Marine Science Plan for Europe that was given priority consideration under the 6th Framework Programme and is being instrumental in defining guidelines for the 7th Framework Programme.

15

Summary of Research and Development Work during the period 2002-2004.

This section summarizes the R&D work undertaken in the scope of the plan set forth in the Associated Laboratory proposal. For details, please refer to the Annual Reports and the Annex. The following major areas were identified:

Dynamical Systems and Marine Robotics

Over the past decade, there has been tremendous interest in the development of advanced robots for operations at sea. Marine robots allow access to otherwise unreachable regions of the ocean and can, in principle, simplify the task of acquiring ocean data fast and cost-effectively without constant supervision of a human operator.

Over the past years, the ISR/IST has been involved in a number of projects that have culminated with the deployment and operation of marine robots at sea. The European MARIUS/SOUV project, coordinated by ISR/IST, witnessed the development of the first civilian AUV in Europe. The European ASIMOV project, also coordinated by ISR-IST, led to the development of advanced systems for the coordinated operation of the INFANTE AUV and the DELFIM ASC, both designed and built in Portugal under the auspices of the FCT. More recently, in the scope of the Portuguese CARAVELA project coordinated by IMAR-DOP/UAzores, the ISR-IST has participated in the development of an autonomous research vessel for long range open ocean operation. The project is a landmark in the development of future ocean platforms that can replace normal research vessels in the more repetitive types of missions. During the past three years, ISR/IST has embarked in the development of two new marine robots: the DREAM ROV, for operations down to 1000 meters, in cooperation with CREMINER and the Faculty of Sciences of the University of Lisbon, and MAYA, a miniaturized AUV for commercial and scientific applications, in cooperation with IMAR-DOP/UAzores and the National Institute of Oceanography, Goa, India.

In spite of the achievements made in the field of marine robotics, much work remains to be done to before such vehicles become ubiquitous instruments in the marine science “toolbox”. Meeting some of the challenges for future vehicle development and for bringing them to bear on the development of faster, cheaper, and more efficient methods for the exploration and exploitation of the ocean, was one the key objectives of the proposed research program that was put together by the Associated Laboratory. This called for a threefold research and development effort that addressed theoretical and practical engineering issues, as well as issues related to the interplay between marine sciences and marine technology. The work carried out led to the study and development of:

Analysis and design tools in the area of navigation, guidance, and control (NGC) with applications to the development of highly performing systems for autonomous air and marine robots. The latter are expected to play an important role in the study of coastal systems (including fast mapping of sand dunes and studying the evolution of their volumetry), as well as in missions that require the coordinated operation of air and surface craft.

Feature-based navigation methods for autonomous underwater vehicles.

Innovative algorithms for adaptive robust control of highly uncertain systems. Strategies for coordinated control of multiple autonomous vehicles that are well rooted in

nonlinear system theory.

Methodologies for air and marine robot modeling and parameter estimation as a view towards the development of integrated plant-controller optimization methods.

Software and hardware architectures for prototypes equipped with real-time Mission Control Systems.

Tools for acoustic and scientific equipment interfacing.

This work is being carried out in cooperation with institutions worldwide. Specially relevant are the cooperation links forged with India, Brasil, Norway, France, and the USA. At the same time, the group pursued the execution of missions at sea to transition from the laboratory to the real world and to foster the symbiosis between marine science and technology. Missions took place in continental Portugal in the areas of Lisbon and Sines, and every summer in the Azores in cooperation with the IMAR/DOP/UAzores using the prototype robots developed at ISR/IST.

16

Ocean Acoustic Tomography, Underwater Communications and Passive Source Localization

During the period reported, the unit of ISR located at the University of the Algarve pursued a research program in the areas of ocean acoustic tomography, underwater communications and passive source localization.

Under project ATOMS (Acoustic Tomography MOnitoring System), work continued on the development of an integrated system for large-scale ocean monitoring, with a special focus on the study of the upwelling filament structure off the Cape São Vicente, one of the most developed and recurrent filaments observed that has important implications in the biological and chemical exchanges between the coastal and offshore ocean. The project includes also an important task on the feasibility of a tomographic acoustic network to monitor the entire Portuguese EEZ. The envisioned network consists of 4/5 acoustic emitters/receivers located in the Azores, continental Portugal and Goringe bank and/or Madeira island. The modeled network will use archived data of temperature/salinity profiles (from NODC and BODC data bases). Studies are under way to determine the main characteristics of the network in order to achieve a given performance in terms of temperature and current resolution.

Under project LOCAPASS (Source Localization with a Random Field of Sonobuoys in shallow water), work progressed on the development of an integrated system for passive source localization using a single or a small number of acoustic sensors. The principle is based on using the environmental information as modeled by an acoustic propagation model to discriminate a perturbation introduced by an active source. Previous results have shown that: i) if the sound source has a relatively large time-bandwidth product and ii) the environment is sufficiently known and stable, it is indeed possible to obtain single sensor source localization in range and depth. This project is now concluded.

Under project TOMPACO (Tomografia Passiva Costiera), the work was focused on the development of a new method for underwater acoustic tomography that uses as source signal the noise emitted by ships of opportunity, received on vertical arrays distributed along the coast, with the objective of measuring the water temperature and currents of oceanographic interest. This project is concluded.

Finally, in the scope of project NUACE (Non-Cooperant Underwater Acoustic Channel Estimation), initiated in early 2004, work is underway on the development and testing of the experimental feasibility of environmental model-based methods to estimate the underwater channel acoustic impulse response. Channel estimation is a common problem to many fields of research and, in particular, in underwater acoustics where the received signal is prone to severe time-space variability, strong multipath, dispersion and reverberation. One of the applications sought is underwater communications, where accurate and updated estimates of the channel response are required for signal equalization, when transmitting over long range shallow water channels.

In 2003, the ISR / unit of the Univ. Algarve participated in a landmark field experiment: the Maritime Rapid Environmental Assessment'03 sea trial, organized by the NATO Undersea Research Centre (NURC), La Spezia (ITALY), from 18-26 June, 2003, off the west coast of Italy, north of Elba I. This participation was made under the AOB-Joint Research Project involving SiPLAB, NURC, ULB (Belgium) and RNLNC (The Netherlands). This participation was naturally continued in 2004, during the Maritime Rapid Environmental Assessment'04 sea trial, organized by the NATO Undersea Research Centre (NURC), La Spezia (ITALY), from 26 March-11 April, 2004, off the west coast of Portugal, near Setubal. The objectives of the MREA sea trials was twofold: the usage of acoustics for rapid environmental assessment of shallow water zones (Acoustic REA) and high-frequency probing the ocean for underwater communications.

Underwater Communications and high-frequency tomography are the scope of another starting collaboration, the High-Frequency initiative (HFi), under which the ISR unit is involved with several organizations, such as SAIC - Scientific Applications Inc. (USA), University of Delaware (USA), SPAWAR (San Diego, USA), University of Washington (USA) and the Naval Research Laboratory (NRL, Washington). A sea trial is foreseen in the Island of Kauai (Hawai-USA), in July 2005.

The results obtained so far in the scope of the above mentioned projects have been presented in international forums and, in some cases, published in peer reviewed journals. See the Annex.

Geology, Volcanology, and Deep Water Hydrothermalim: major achievements

Hydrothermal activity in the Azores area (MOMAR area)One of the major achievements of the teams involved in studies of deep water hydrothermalism (CREMINER and IMAR-DOP/UAzores) was cruise Seahma-1, in July-August 2002, using the largest French oceanographic ship, l’Atalante, and the new and highly sophisticated ROV Victor 6000. Victor 6000 was at

17

the bottom of the ocean for a total of 176 hours, for geological, ecological and hydrothermal mapping, and for various types of sampling. These included biological specimens from hydrothermal vent communities and surrounding areas (fish, crustaceans, gastropods, microbes and many others rocks and sediments and hydrothermal fluids from high temperature chimneys and almost cold discharge through sediments). Hundreds of specimens were collected and are still under study. IMAR-DOP/UAzores deployed four cages for fauna, partly filled with mussels (that were retrieved during the winter), in short expeditions made possible by the immediate availability of the local oceanographic ship Arquipélago. A large part of the Seahma-1 operations took place on seamount Saldanha, which hosts a subtle but important hydrothermal field discovered in 1998 by cruise Saldanha. Extensive geological and ecological mappings were done around Mount Saldanha, with some surprising results, including previously unsuspected variations in faunal abundances. Significant advances were made in the process of understanding mineral deposition, on the Rainbow and Lucky Strike fields, since it was possible to collect new materials, from fault scarps, using not only Victor 6000 but also a combination of precision dredging and subsequent inspection with the submersible. This permitted detailed mapping of the products collected with the dredge. Using these techniques, new evidence was produced in favour of sub-seafloor mineral precipitation.

In 2003, work continued in the framework of project Seahma hindered by difficulties in obtaining ships and geophysical equipment for the Seahma-2 cruise at sea, which was mainly dedicated to geophysics. The cruise had to be postponed to 2004. Meanwhile, CREMINER was actively involved, in close co-operation with the partners at IMAR-DOP/UAzores, in efforts to maintain and develop national and international interest in assembling a deep-sea observatory in the MOMAR area. This resulted in international agreements to develop an Eurocores project (under the auspices of the European Science Foundation). Nationally, much effort was dedicated to the subject in the framework of a Working Group on Deep Sea Research (Grupo de Trabalho em Investigação em Mar Profundo, GT-IMP) created in the framework of the Comissão Oceanográfica Intersectorial (Intersectorial Oceanographic Commission). As a result, the concept of a permanent observatory is gaining supporters and there are reasons to believe that the 7th Framework Program (EU) will include provisions for this.

Deep Sea research is also active in the field of gas hydrates – Creminer is beginning to participate in this field as well, with a perspective related with the methane-producing hydrothermal fields such as Rainbow and Saldanha. Also of great interest is the preparation of a workshop on the strategic development of deep sea research in Portugal. These aspects (MOMAR, Gas Hydrates and Deep Sea Research in general) have been included in the documents produced by the National Strategic Commission on the Oceans (launched in July 2003).

Exploration of the Serreta submarine volcano (near Terceira Island, Azores)

In 2002, a cruise was conducted to the site of the Serreta submarine eruption (1998-2000), to investigate the distribution of volcanic products over the area. Several dredges and grabs were collected, down to a depth of 1600 metres, which permitted the storage of fresh, well calibrated volcanic material, most certainly from the 1999 eruption.

Ocean Drilling Program (ODP) Leg 193

The Ocean Drilling Program Leg 193 Second Post Cruise Meeting took place from 14-19 October 2002, in Lisbon, Castro Verde and Aljustrel, with participation of 18 researchers from Australia, Germany, Japan, Papua New Guinea, South Korea, United Kingdom, USA and Portugal. The purpose of the meeting was reciprocal reporting on post-cruise research. ODP Leg 193 took place from 7 Nov 2000 to 3 Jan 2001, in Papua New Guinea waters and was dedicated to study of the Pacmanus active hydrothermal field in a back-arc basin. Two Creminer researchers participated in the leg (Fernando Barriga, Co-Chief Scientist, Alvaro Pinto, Ore Petrologist). The meeting took place in Portugal by unanimous decision of the scientific participants, given favourable logistics and also the chance of including technical visits to mines in southern Portugal (Neves Corvo and Aljustrel), geological analogues of the Pacmanus active system. In 2003, Creminer continued to be actively involved in ODP/IODP, not only through publications related to ODP Leg 193 but through participation in ESSAC (European Science Support & Advisory Committee).

The Prestige sinking and oil spill

In the framework of contractual obligations as part of the Associated Laboratory, Creminer paid maximal attention to the Prestige sinking and oil spill, given its expertise in deep sea science. News were follows with great attention and there was a strong commitment to the participation in workshops on the case (Faculty of Sciences, Lisbon and University of Porto, Faculty of Engineering – in January and May, respectively) analyzing the mitigation proposals and testing the hypothesis that deep sea hydrothermal

18

fauna (especially microbes) could be useful to destroy the Prestige fuel oil, given its high sulphur content. This is being done in the framework of project Seahma. Final results are not yet available.

On a different note, Creminer is actively involved in exchanges with Japan, in the areas of deep sea research (deep biosphere and observatories) and on a more remote connection, that of the study and remediation of sediment trapping in river dam reservoirs. Creminer research has shown that sediments accumulated in the bottom of reservoirs can be extracted and used as soil fertilizers; future research will also exploit the possibility of transferring the coarse fraction of such sediments to the coast (directly or indirectly). The UNESCO Geological Correlation Programme has approved the project. Participating countries are Portugal, Brazil, Canada, Japan and China.

From coastal waters to the deep-sea environment: ecosystem studies and oceanography

The research program proposed targets the Azores as a natural laboratory for the study of a vast number of challenging scientific issues in marine science. In this context, the IMAR-DOP/UAzores has placed considerable emphasis on studies of ecosystems and oceanography, from coastal waters to the deep-sea environment. The IMAR-DOP/UAzores is a member and part of the Steering Committee of the Network of Excellence (NeO) MARBEF - Marine Biodiversity and Ecosystem Functioning, which includes Coastal areas, offshore environments, and the deep-sea environment. This NeO is a VI Framework program financed by the EU. The NeO aims at integrating research efforts by forming a dedicated group of marine scientists and institutes and creating a virtual European institute with a long-term research programme and dedicated links with industry and the public at large. Better integration of research is also required to support the legal obligations of the EU and its member states and associated states for the Convention for Biological Diversity, the OSPAR and Barcelona conventions as well as several EU directives (Bird Directive, Habitat Directive, and Water Framework Directive). This motivates the reporting of the work carried out during the period from 2002-2004 according to the following main themes.

Conservation of Marine Habitats and Biodiversity

Located in the middle-Atlantic Ridge area, the Azores consists of nine islands, several islets, more than 100 seamounts and a yet unknown number of other deep-sea habitats. The recent increase of human activity has been affecting the marine environment. The IMAR-DOP/UAzores has been involved on the preparation and execution of all scientific activities leading to the elaboration of management plans for the 15 marine Sites of Conservation Importance and the 13 marine Special Protection Areas declared for the Azores under the Habitat and Bird Directives of Natura 2000. This was done under the scope of the LIFE Project – MARÉ and Interreg IIIb Project – OGAMP, both funded by the European Commission. Special attention was given to the conservation and valorization of marine biodiversity (including sea-birds and marine mammals). Further to these there has been much involvement on the gathering and reviewing of scientific information in view to sustain proposals and management plans for other areas that can be considered under the scope of the Annex V of the OSPAR convention, namely, the deep sea vent sites Lucky Strike and Menez Gwen and the banks and seamounts of Formigas/Dollabarat, D. João de Castro and Sedlo. This last one is the central target of the EC FP5 project OASIS.

Both in coastal and deeper areas, it was necessary to use advanced tools for the development of detailed maps that are instrumental in deriving guidelines for marine management. The maps can be used to determine habitat attributes that significantly affect the distribution and dynamics of species and to monitor communities’ structure and species abundance. This work was developed under projects MAROV (funded by the PDCTM), and is advancing through the MAYA Project (funded by National Government through ADI), and Submarine Growth of Volcanic Islands Project (funded by IMAR and other British institutions). These projects are highly multidisciplinary, and display a clear symbiosis between marine science and technology. They integrate key components of study and give support to modern systems for marine habitat mapping: data acquisition, pre-processing, and transmission, multi-scale data integration, and information processing and dissemination. The areas around S. Jorge, Faial and Pico islands are at the moment mapped with high resolution. The dominant habitats and new clues about the sedimentary dynamic of the shallow water seafloor are identified.

To test and evaluate the benefits of marine protected areas another two projects were initiated in 2002.These projects, MAREFISH (funded by FCT/MCES) and TelAzo (funded by the Luso-American Foundation), have as main objectives the experimental evaluation of benefits of Marine Protected Areas, using fish as case studies and telemetry as main tools. With this tools it can be now clearly defined the daily and seasonal dynamic of two important commercial fish species. These results are essential to define the marine protected areas borders.

19

Another important aspect in conservation and habitat mapping is the work developed by the cetacean team. The CETAMARH Project (funded by FCT/MCES) intends to add to the diversity of ecological and social behaviors currently known, and provides the opportunity for testing hypothesis concerning the ecological determinants of cetacean distribution, ecology and social structure. The role and importance of CETAMARH was recently reinforced by the approval of MACETUS, an Interreg IIIb funded Project.

Offshore Environment

The main scientific challenge in the study of the offshore environment undertaken at the IMAR-DOP/UAzores is the seamount studies. IMAR-DOP/UAzores is also involved in the European OASIS project. This project aims at describing the functioning characteristics of seamount ecosystems. Several missions were already preformed around the seamounts located farther away off the Azores.

The oceanographic team of the IMAR-DOP/UAzores has also been involved in several other projects like DETRA (funded by the Regional Government); AERONET (funded by NASA). Under the scope of these projects SST patterns, variability and associated dominant space and time scales in the Azores (including the Azores Front and Azores Current) have been defined. These investigations emphasize the dual influence of the Gulf Stream and the Azores Current in this region. The SST anomaly field shows that the Mid-Atlantic ridge forms a barrier for heat exchange between the western and the eastern subtropical North Atlantic basins finally; the impact of North Atlantic Oscillation (NAO) forcing on SST distribution has been demonstrated.MARECO Project is a pilot project within the Census of Marine Life programme; an international research programme exploring the abundance, distribution and diversity of life in the world's oceans. This project aims the study of the patterns and processes of the ecosystems of the northern mid-Atlantic.

Deep-Sea Ecosystems

IMAR-DOP/UAç is deeply involved in the study of deep sea hydrothermal vents. These recently discovered and fascinating areas are now the focus of interesting and challenging studies that have received wide attention in Europe and worldwide. Research work in this exciting field is being carried out in the scope of the VENTOX and the PDCTM SEAHMA Projects.

The VENTOX project (funded by 5th FP of EC) was intended to provide innovative insights into the long-term adaptations of organisms from the deep-sea hydrothermal fauna which are exposed to natural high levels of environmental contamination. Within Ventox project, IMAR-DOP/UAzores has developed the LabHorta concept that allowed for the on-shore study of deep-sea living animals. To achieve this, several cages were placed near the vents by the submersible Nautile and the ROV Victor. These cages were retrieved, by remote control, and moved to LabHorta using the R/V “Arquipélago” according to the supplying needs. This operation allowed year round availability of living organisms. The main tasks of IMAR-DOP/UAzores in this project were the study of ecotoxicology, genetics and trophic relations organisms. The scope of SEAHMA Project (funded by PDCTM) includes the study of global biogeochemical cycles with special attention being paid to the recently discovered Saldanha hydrothermal vent field. The main objective of IMAR-DOP/UAzores is the study of energy fluxes and contaminants in hydrothermal vent food webs. VENTOX project finished early 2003, however, LabHorta kept running under the SEAHMA project, and several cooperation links were established among international scientific teams. During the SEAHMA project, it was possible to determine the reproductive season of the deep-sea hydrothermal vent mussels, when the January retrievable cages mussels presented gonads on the ripe state. Moreover, the experiments that are being carried on in LabHorta, allowed the discover that the captivity mussels kept for months in controlled aquarium conditions, developed gonads into a ripe state on the same season. Also the possibility of keeping mussels alive in good condition allowed the studied of endosymbiotic bacteria transmission on Bathymodiolus azoricus, and the behavioural response of the mussel to exposure to contaminants. In Late 2003, two FP6 projects were under negotiation to be contracted in 2004. The STREP EXOCET (Extreme ecosystem studies in the deep ocean: Technological developments) and the MOMARNET Marie Curie network. These two projects are important steps of the MoMAR concept. This team has been very committed, and was involved in the organization of several meetings like the MoMAR meeting in Paris (March 2002); Lisbon (December 2002) and in Barcelona (September 2003) and the workshop in Faial (June 2002).

During the call for EoI (Expressions of Interest), the IMAR-DOP/UAzores coordinated an EoI-IP (Integrated Project) names “Long Term Monitoring of the North Atlantic: an Observatory Approach,” that received the support of 31 leading institutions in Europe and small and medium enterprises.

IMAR-DOP/UAzores is strongly involved in activities of out-reach and education related to the abovementioned areas, as detailed in the Annex.

20

Marine Science and Technology: operational support.

The Azores, with one of the most extensive EEZ of Europe (1 million square Km) is located in the Mid Atlantic Ridge (MAR) in a highly active geological, geophysical and oceanographic area, where some of the most dynamic ecosystems of the planet, such as seamounts and deep-sea hydrothermal vent systems occur. The MAR near the Azores is ideally located for marine multidisciplinary studies: it is near port, allowing for short transit times for the deployment and retrieval of tools and, even, live organisms. The research and development efforts proposed by the Associated Laboratory address important practical applications in the fields of marine biology, geology, and oceanography, the main core of the missions envisioned taking place in the Azores region. Two main ecosystems are being studied by the Associated Laboratory: Coastal environment and deep-sea environment.Throughout this research endeavour, the R/V “Arquipelago” and the smaller ship “Águas Vivas” have become extremely important infrastructures used by the different teams of the Associated Laboratory. Both have consistently served as support vessels to a series of missions involving AUVs (namely, the ARIES from the Naval Post Graduate School of Monterey, USA and the INFANTE that is property of ISR/IST) and the DELFIM Autonomous Surface Craft (operated by ISR/IST). The R/V “Arquipélago” was involved in several missions to recover the deep-sea retrieval cages at Menez Gwen and fisheries at Lucky Strike, under the scope of VENTOX and SEAHMA. This ship was also adapted in the summer of 2003 to accommodate a MultiBeam Sonar of the University of Cardiff with the objective of studying the shelves of the islands of Faial, Pico and S. Jorge.In 2002 and 2004, the Arquipélago was used extensively by IMAR-DOP/UAzores and the IGM (National Geological Survey) in the mapping of sand banks around Pico and S. Miguel in the scope of projects GEMAS and MARINOVA. For this task, the research vessel was equipped with a CHIRP acoustic and a BOOMER. The vessel was further used to for dredging with corers during the same projects.In 2002, the Arquipélago was used in a mission to the underwater volcano of Serreta in the scope of project STAMINA of the University of Lisbon. This operation will be continued in the second half of 2004. Overall, the Arquipélago is an extremely valuable platform, capable of providing services to both national and international institutions. This more than justifies the need to re-equip the vessel to meet the ever increasing demands of marine scientists.

Work done. An evaluation of the deliverables in the original proposal.

Four main deliverables were mentioned in the original proposal of the Associated Laboratory for its first 5 years of existence. Tight budgetary constraints have placed serious limitations on the scope of the work that could be accomplished. In spite of these limitations, however, the units involved in the Associated Laboratory managed to push the main topics set forth in the proposal and in some cases to make substantial progress, at the expense of a considerable burden on the financial side.

D1. Habitat mapping of selected marine areas of economic and scientific interest in the Azores. Considerable progress was made towards the development of an integrated data acquisition, transmission, and information processing system that relies on the operation of autonomous marine vehicles equipped with scientific sensors, followed by information fusion and information dissemination using a Geographical Information System (GIS). So far, ISR/IST and IMAR-DOP/UAzores have successfully used surface ships and the marine robots DELFIM and INFANTE for automatic bathymetry operations using a side-scan unit, a mechanically scanning pencil beam sonar, and a multibeam sonar. This was instrumental in obtaining high resolution maps of the seabed around the islands of Faial, Pico, and São Jorge. The main bulk of the work was carried out in the scope of the MAROV project of the FCT in cooperation with IG (the Geological Survey of Portugal).

D2. Observation of the deep and shallow ocean floor in the Azores using marine robots and vision / acoustic systems. The partner group is currently cooperating on the development of a tele-operated vehicle (ROV) capable of diving to the depth of 1000 meters, in the scope of the DREAM project of the FCT. Emphasis is being placed on equipping the vehicle with advanced vision systems for seabed visual reconstruction and seabed classification, as well as automatic systems for precise trajectory tracking, path following, and target hovering. Foreseen applications include time-series studies (biology and geology) on the Menez Gwen hydrothermal site (Mid-Atlantic Ridge, Portuguese EEZ, depth 800 meters). Sea trials are planned for 2005.

D3. Observation of shallow water and coastal environments using an autonomous air vehicle. Over the past few years, as a natural consequence of a longstanding collaboration program with the Department of Aeronautics and Astronautics of the Naval Postgraduate School of Monterey, California, USA, ISR/IST has started to apply some of the methodologies and technologies developed for ocean

21

vehicles to the control of air robots (helicopters). This is justified in view of the increasing interest worldwide in the use of unmanned aerial robotic vehicles to perform airborne surveying tasks, including fast mapping of sand dunes. As part of this effort, ISR/IST has finished instrumenting an unmanned robotic helicopter. The platform is based on an industrial radio controlled helicopter that was equipped with a distributed real time computing network, a reliable wireless communication system, and sensing devices. A first series of field tests was completed in early 2004, with the helicopter being controlled via a radio link. The sensor data obtained with the instrumentation installed on-board is now being used for model parameter estimation and to assess possible problems caused by excessive vibrations. The activity pursued in this area is well rooted in scientific applications that require the use of autonomous air robots to accurately map coastal areas subjected to erosion, using airborne laser altimetry. In particular, project ALTICOPTER funded by the FCT envisions the use of a helicopter to map sand dunes along the Portuguese coast.

D4. Installation of a shallow/deep water online ocean monitoring system. This and related objectives were tackled by ISR / Univ. of Algarve in the scope of a number of projects that include ATOMS and TOMPACO, described before. The final envisioned goal is the reconstruction the 4D evolution of temperature and oceanic currents by using remote sensing acoustic signal transmitted and received on fixed stations over an area of interest (on the order or up to 100 square kms). Relevant theoretical results have been obtained and published in reputed conference proceedings and journals.

22

3.2.Thematic Area B: Monitoring and Surveillance using Robotic Agents

The strategic objectives of Thematic Area B remain fundamentally the same as planned in 2002. The last two and half years of work have allowed to achieve progress in most of the areas, particularly in all aspects related to the basic disciplines. Regarding the advances in the application areas, important progress has also been attained but somewhat undermined by the lack of resources.

The application areas require a larger effort in integrating individual modules or algorithms into function (partially or fully integrated) prototypes. As such, it was planned that several post-docs researchers would be involved in these tasks, directly or coordinating younger researchers or undergraduate students. Since funding delays have precluded hiring such post-doc researchers, the level of integration achieved is not as advanced as planned.

In addition, the application areas require the existence of specific equipment for deploying the developed methodologies. For that reason, a major proposal was prepared for the F.C.T (“Programa Nacional de Re-equipamento Científico”) in 2002. The result of its evaluation is still unknown.

As a consequence of the scarceness of human resources and equipment, a significant effort was developed for raising the necessary funds to advance research in this Thematic Area. Researchers have been particularly aggressive in applying to EU, FCT (very recently), ESA (European Space Agency) and ADI (Agência de Inovação) project funding programmes. This effort has allowed to raise the necessary research resources at the cost of dispersing the activities more than what was envisaged initially.

We expect that, having the possibility of hiring post-docs researchers, as well as the ability to purchase the necessary research equipment, will be essential to fully address all the aspects described in the original plan in a fully satisfying manner.

Scenarios of interest

The needs for monitoring and surveillance has increased enormously in the near past, motivated by the wide range of application scenarios that can be envisaged and by the level of maturity reached by the main enabling technologies.

On the other hand autonomy allows for extended unsupervised operation as well as for decentralised decision-making. Networking of multiple, low complexity autonomous systems meet spatial sampling requirements and allows for dynamically reconfigurable topologies to match the dynamic properties of the involved phenomena. Recent advances in Complex System Theory and Signal Processing, coupled with the advent of miniaturized sensors and actuators, powerful embedded processors, and wireless communication systems, have afforded engineers with the methodologies and technologies to actually build networks of autonomous systems with different functionalities and a large scope of applications. These include operations in hazardous environments (ocean, space, contaminated areas, areas destroyed by natural disasters, etc), in industrial or civil engineering structures (pipeline monitoring and surveillance, bridges, dams), and in services (buildings, public areas, traffic monitoring and surveillance).

A set of enabling methodologies required to handle this class of problems includes: Robotics; Artificial Intelligence; statistical Signal Processing and Computer Vision; Control Theory; Distributed Real-Time Computing; Operations Management; and Optimisation.

A number of robotic surveillance and monitoring application domains where the Associated Laboratory has potential to be an active player in the future, namely Space Applications, Large-scale Land/Aerial applications and Indoors/Outdoors scenarios, are described in the sequel.

Space Applications – The recent association of Portugal to the European Space Agency has raised both the scientific and economic interests in spatial applications. For example, the Aurora programme, currently under preparation in the context of the ESA, aims at formulating a long-term strategy for the robotic and human exploration of the solar system bodies. Such programmes will undoubtedly open new opportunities for European countries at political (exploration of solar system), scientific (solar system evolution, life origin, human expansion/exploration to/of other planets) and technological (the European Space Agency has identified the following enabling technologies: robotics, AI, telecommunications, energy, among others) levels. From the strategic point of view, this is a challenging opportunity for

23

Portuguese R&D institutions. In many cases, such a challenge will require the development of new sensors and manipulation systems, automatic docking of space structures, landing systems, tele-operation (with controlled autonomy) of robotic systems over extremely large distances, telepresence capabilities, distributed systems, etc.

Large-scale land/aerial applications – Monitoring and surveillance of large-scale areas (e.g. forest, agricultural and coastal regions, as well as urban areas) are of the utmost importance both from the economic and environmental point of views. On the social aspects, one can add applications like search and rescue, traffic surveillance, fire detection, to name just a few. The methodologies for distributed sensing and monitoring may be applied in the long-term future. Forest fire, for example, has caused an economical loss estimated in 90 million thousand escudos in the year 2000 (DGF – Direcção Geral das Florestas) for a total burnt area of 160 thousand hectares. The nature of these applications often precludes the use of satellite imagery alone but rather, they require the coherent and co-ordinated intervention of a distributed network of sensors and/or robots operating both at land and air.

Indoors/outdoors scenarios – The impressive proliferation, performance boost and cost reduction of computers, cameras, internet access and mobile communications has created the opportunity for new mass-markets at the levels of domestic or even personal applications. In indoors scenarios one can emphasise multimedia or “attentive” spaces capable of perceiving the human actions and reacting accordingly (e.g., for offices, education or entertainment). Another key demand is that of providing a better assistance and monitoring to the ever-growing number of elderly (and handicapped) citizens. In Portugal, assistance to the elderly has the cost of about 400.000 Escudos per active person per year – www.ine.pt - and 15% of the population has 65 years or more. These numbers are expected to rise dramatically in the forthcoming 10-25 years, posing new social, economic and cultural problems. Robotic technologies can thus provide the means for monitoring elderly citizens as well as providing friendly and safe interfaces to a variety of domestic appliances and systems. Most of the same rationale applies to children and infants. Other applications include monitoring open spaces like schools, sports facilities, malls, airports, as well as urban traffic.

Rationale

Enabling Technologies

Although apparently very diverse, the application fields mentioned in Section 1 are grounded on a number of enabling technologies, where ISR-Lisbon has recognised expertise and which can be further developed by focusing in the needs associated to these application scenarios. In particular we can refer the following enabling technologies and give examples on how they could be developed under this thematic area.

Computer Vision

In many of these scenarios vision is an extremely rich source of information about the environment. Binocular or monocular tracking systems can be used to track people or other moving objects, identify some pattern of activity (behaviour) and detect unexpected occurrences (e.g. traffic monitoring or surveillance). Vision can also be used to track people’s faces, interpret facial and body gestures and become the building block of novel multimodal interfaces to new appliances and systems. Also, future generations of multimedia systems, digital libraries and image communication, will borrow from computer vision the crucial techniques for complex tasks like image indexing, human-machine interface or interactive video.

Methodologies for 3D reconstruction have matured to the point where 3D models can reliably be computed from image sequences and used for recognition, augmented reality, provide information to elderly or handicapped, create CAD models for manufacturing, build spatial representations of objects or surfaces, video mosaics, etc.

Similarly to many living beings, the dynamic visual information contained in a video stream can be used for controlling a robot evolution in a certain environment either in land, air or space applications. For years, ISR has worked on vision-based navigation systems, often inspired after biological findings, and has the experience of developing such systems for land air or underwater vehicles. New vision systems (e.g. catadioptric panoramic cameras) can be utilised to match the exact needs of special applications.

Robot Navigation

24

Robot navigation and map building will be necessary for many of these applications. Robot navigation provides controlled mobility to the sensing devices required in the referred applications and environment mapping supports navigation and is the expected outcome in some applications.

The kinematic structures, locomotion capabilities, onboard sensing, cost, computational power, learning/adaptation are all parameters that need to be evaluated simultaneously when addressing some of the above mentioned applications. There are fundamental questions that must be addressed both for understanding the problem of navigation itself and for developing reliable navigation systems/robots based on a limited computing and sensing capacity able to interact with humans and other robots in a productive manner. What representations should be built in order to map the robot’s environment? How can user-adaptation take place beyond pure programming? How can navigation take advantage of distributed sensing and locomotion capabilities in a fleet of robots?

Distributed AI

In a co-operative robotics context, which is the one involved in many of these scenarios, in order to obtain useful co-operation, relationships among robots have to be accomplished. A group of agents must execute joint work to solve a particular problem, mainly due to the existence of dependencies between agents’ actions, the definition of global constraints and goals, and the fact that each individual might have insufficient competence, resources and information to solve the problem per se. For uncertain, complex and dynamic environments, a coordination-based approach is doomed to fail since it is almost impossible to anticipate and pre-plan for all possible contingencies.

Therefore, the idea of developing general methodologies for teamwork in a multi-agent system seems to be a more natural, flexible and efficient way to solve cooperative problems. The goal is to create a formal modelling tool adequate to develop and organise a team capable of dealing with complex and dynamic environments, working coherently as a group of agents, handling different and even opposite views of the world and the problem within the team, allowing flexible communication among team members, evaluating the team performance and implementing re-organisation strategies to handle unexpected situations. In order to develop such a tool, a significant number of issues must be addressed: agent architectures (e.g., logic-based, reactive, BDI, hierarchical), methodologies for establishing and terminating cooperative actions (e.g., Joint Intentions, Shared Plans), teamwork communication protocols, re-planning in order to recover from failures, and multi-agent systems learning.

Control

The usage of non-linear multivariable control systems based on modern control techniques is of utmost importance for the correct operation of the type of robotic devices envisaged for the large scale land/aerial and space applications, such as outdoors land vehicles, semi-autonomous helicopters, automatic docking, to name but a few.

Large scale systems composed of several robots can be seen as distributed embedded systems, defined as devices used to control, monitor or assist the operation of equipment, machinery or a plant, where the term "embedded" reflects the fact that the devices are an integral part of the system. The overall integration and coordination of the different subsystems that compose large scale systems requires mixing continuous state-space time-driven with discrete state-space event-driven system models and control techniques. Therefore, Discrete Event Dynamic Systems and Hybrid Systems are promising tools to carry out the analysis and design of multi-agent systems such as networks of land/air robots or fire detection stations.

Application Areas

ISR-Lisbon has identified a number of areas from the general set described in Section 1, where its expertise in the listed enabling technologies can be applied in the upcoming 10 years:

Search and Rescue: In the context of Air/Land applications, Search and Rescue (SAR) robots with a considerable degeree of autonomy can be used in hazardous scenarios, like those arising from earthquakes, floods or environmental pollution. Collaborative links have been established with the Sistema Nacional de Protecção Civil (SNPC) and others may be established in the future with researchers on related scientific areas, like Civil Engineering, Geology or Geographical Information Systems (GIS). The long-term goal consists of making available large fleets of heterogeneous robots, endowed with means for detection and rescue of victims. Besides the utilization of available information from GIS, this (multi-robot)

25

system could be used in a coordinated fashion with simulation models of a natural catastrophe (e.g., earthquakes) developed or being developed by the Lisbon Mayor, the SNPC and other institutions, to predict the impact of disasters for a time window after they actually occur. This strategic plan naturally promotes basic research on the enabling technologies and knowledge reinforcement on more specific areas:

Outdoors multi-sensor navigation

Multi-agent systems

Task coordination

Multi-sensor aerial 3D-reconstruction.

Control of complex dynamical systems

Surveillance of Urban Areas: Urban spaces are increasingly suffering from the problem of traffic density. Traffic lines depend not only on vehicle density but also on the overall organisation of traffic flow and on some particular “hot” points with high concentration of vehicles. With the ability of monitoring and gathering information on such hot points, adequate traffic control policies may be designed with a significant impact on the citizen’s well being and reduction of polluting emissions. Monitoring and surveillance technology can also be used to fight crime, violence and vandalism that tend to increase in large highly populated cities. Particular care can be devoted to public spaces like schools, transports, stadiums and degraded neighbourhoods.

Robotic Assistance: Handicapped or elderly people life quality at home can be considerably improved with the assistance of robotic devices, to provide food, pick up and transport objects, etc. ISR has a long-standing experience in vision based control systems, such as binocular tracking systems and vision based navigation. This expertise has been applied in various contexts from land vehicles to flying (an aerial blimp) and underwater robots, where visual tasks have been successfully experimented at sea. The combination of binocular tracking systems and robotic manipulators can be used as a first step towards the full development of a robotic assistant that can be very helpful in indoors applications referred above. One of the long term goals is that of endowing the system with learning and development capabilities. The control system design approach will enable the system learn from its own experience (e.g. how to optimise the various head/arm controllers, visuo-motor coordination, object representation, etc) rather than full programming of the system functionality, as in current approaches.In the future, these systems can be applied either in space applications, where self-calibration, flexibility and learning possibilities are needed to extend the system’s utility and time of usage. Similarly, for interacting with humans (e.g. elderly assistance) it is of the utmost importance that the system can adapt to and learn from each specific user and recognize the specific user’s gestures and goals. The operation in real complex environments is far too complex and rich to imagine that a robot could possibly be pre-programmed to deal with all possible situations that may arise. In the long-term, a robotic assistant can be seen as part of a network of sensing devices (mainly vision cameras) spread around the house, which anticipate and/or register requests from people in different rooms and communicate the corresponding required tasks to the robot.

Actions Proposed. Deliverables

D1 - Search and Rescue

Year 5

In a 5 year horizon, we propose to present a prototype of a robotic system composed by a small number of agents, namely one land and one or two air vehicles able to perform cooperatively a set of relatively simple tasks:

One aerial robot acquires data of the environmental map of the area in order to inform and assist the navigation of the land vehicle and possibly the other aerial robot towards particular regions, which might be the most critical ones in a rescue application.

26

The land vehicle performs simple outdoors navigation tasks on different types of terrain compatible with its locomotion capabilities, and under different environmental conditions (e.g., obstacle density, light conditions).

One aerial robot performs elementary actions, such as flying towards a pre-defined spot, hovering or keep tracking of a spot on land.

year 10

The activities to be carried out in a 10 years horizon represent an extension of the first 5 years activities towards multi-robot systems with a larger degree of autonomy and of cooperative action:

The aerial and the land robots cooperate to build a detailed map of the scene under analysis, as well as to help navigation towards a land spot identified as relevant. Both will rely on outdoor multi-sensor navigation techniques. Coherent environment representations must be maintained by the 2- or 3-robot teams.

One aerial robot flies to the identified relevant location and builds a detailed 3D map using several sensors (e.g., laser, vision) and sensor fusion techniques. This map will be used by the land robot upon its arrival to the location.

The above multi-robot tasks will be designed based on hybrid systems and logic-based teamwork techniques, in order to guarantee both qualitative (e.g., avoidance of unsafe states and/or deadlocks) and quantitative (e.g., reducing travelling time, increasing task reliability) performance, as well as to guarantee logical teamwork consistency.

D2 - Surveillance of urban areas

year 5

We propose to develop monitoring algorithms for the surveillance of urban areas, allowing the interpretation of the video sequences captured by a small number of cooperating cameras, the characterisation of typical traffic patterns as well as tracking of moving objects. This may incorporate agile cameras (i.e. monocular or binocular camera heads), or specially designed catadioptric panoramic cameras. This approach will be a useful tool both for monitoring and security purposes.

The following items will be considered:

tracking of moving objects in video sequences (e.g., persons and vehicles)

tracking of object groups in the presence of occlusions

identification and detection of behaviours

data fusion using multiple cameras

geometric representation in a map

build statistical information of the motion patterns

special purpose (e.g. catadioptric panoramic) cameras may be used.

D3 - ROBOTIC assistant

Year 5

27

We propose to develop a binocular head system incorporating inertial sensors for compensating motion disturbances, thus improving the quality of visual tracking and stabilisation. The head may be attached to a robotic arm, in a humanoid-like arrangement, for short-range manipulation tasks. This system will have the possibility of being remotely operated with high-level task descriptions.

Year 10

We propose to develop methodologies for learning and development in artificial systems, similar to those found in infants, in particular to improve visuo-motor coordination. Adaptation, development and learning from own-experience and from observing other robots/humans may be the key for building truly flexible and autonomous systems. The approach will address the following topics:

Head-eye control and coordination

Understanding of human action/gestures

Developmental approach for overall system design

Attention mechanisms for complexity management

Adaptation to the user and environment

28

3.3.Thematic Area C: Sustainable Technologies and Environmental Systems

Sustainable Technologies are engineering systems designed to promote innovation in harmony with the environment.

Introduction - Research Goal:

The strategic objective, of this thematic area, is to undertake multidisciplinary research aiming at developing and evaluating emerging and alternative complex engineering systems promoting sustainability, namely in terms of the needs to secure socio-economic development and the quality of the environment, creating a more prosperous and sustainable society.

Enabling technologies are being developed and assessed under a systems view, comprising the use and environmental implications of materials, energy, and products in modern societies. To achieve these objectives, research and development activities include the analysis of advanced systems, but also the analysis of the global carbon bio-geochemical cycle and of materials flows in the economy, including product and material life cycle management through reuse, remanufacturing, and recycling.

Main research areas include: Eco-design for sustainability in industry and domestic applications Advanced integrated combustion and thermal systems Total life-cycle energy chain and environmental impact assessment Carbon cycle Flows of materials in the economy from raw-materials extraction to final integration in the natural

environment Techno-economic assessment, technological change and systems integration

Rationale:Current environmental policies have a firm and necessary regulatory foundation, but future initiatives need to be more cost-effective. In this context, technological innovation is the most attractive response to our environmental problems. As a consequence, innovation-friendly environmental policies will require increased research efforts and incentives to environmental entrepreneurs, which should be implemented in order to rely on market-based solutions and provide public information to gauge the environmental performance of economic actors. In short, they will encourage more rapid evolution in the industrial ecosystem, requiring the effective development and implementation of sustainable technologies. This is a long-term project whose success depends on realistic, patient, adaptive, but also persistent visionaries.

The proposed research effort is based on the following specific aspects:

a) Eco-design for sustainability in industry and domestic applications

The economic success of firms depends on their ability to identify the needs of customers and to quickly create products that meet these needs and can be produced at low cost, but also following proper environmental limitations. The approach to be adopted is based on the concept of life cycle, as it considers manufacturing (including raw material processing), utilisation, disposal, re-utilisation and recycling strategies. This requires a multidisciplinary approach taking into consideration that a product is engineered, and therefore product development combines new engineering solutions, product design, marketing strategies and environmental considerations, amongst others. Main applications areas will include: Car components Industry; Consumer Goods; Domotics; Burning equipments.

b) Advanced integrated combustion and thermal systems

The increased restrictions on CO, CO2 and NOx emissions, the excessive weight of energy consumption costs on overall operation costs in industry, and the increased competitiveness of the global marketplace, have promoted the need for innovative solutions applied to combustion and thermal systems, including optimum burning-systems design and the development of combustion control strategies. In this context, the use of lean combustion conditions, as a mean to achieve ultra low NOx emissions, have gained renewed and detailed attention, together with induced external oscillations on the air-fuel line feed.

29

The research program will focus on clean combustion control strategies, emphasizing the following topics: i) Controlled direct injection of liquid fuels for engine applications; ii) Actively controlled flame-flow interaction processes supported by the development of active control methodologies (examples including acoustic excitation, controlled shear strain rate through induced counter-current shear flows, controlled swirl levels-vortex breakdown phenomena in a open or closed loop control); iii) Flame-wall interaction processes, assisted by the potential use of induced oscillations and swirl, as mean to achieve more efficient heat transfer to the wall with controlled pollutant emission; iv) Development of active/passive control strategies applied to incinerators coupled/decoupled from the acoustics of the combustor cavity, which should take into account possible changes in the dynamics of the system to be controlled: besides devising adequate control methodologies for combustion systems, the research will also focus on supervision control systems as a mean to adapt the controller behaviour to the current system dynamics, using intelligent control tools (e.g., Fuzzy Control, Neural Networks, Hybrid Control Systems); v) Optimization of burners design using Artificial Intelligence techniques. Current use burners have been empirically designed through an ungrounded set of trial-and-error tests. As the burners design problem can be formulated as a search problem in a multi-dimensional design space, the application of AI methods (e.g., Constraint Satisfaction, Heuristic Search, Distributed Search) will eventually improve the burners performance measured in terms of heat transfer, pollution rate and burning eficiency.

c) Total life-cycle energy chain and environmental impact assessment

The use of input-output analysis combined with emerging environmental impact assessment tools has become critical to estimate the expected increase in the GHGs in the coming years. The use of the input-output framework will bring the major advantage of considering all the direct and indirect economic effects and avoiding the settlement of arbitrary boundaries. The continuous GHGs evaluation will allow identifying the relative position of Portugal in terms of the Kyoto protocol, which is of critical importance to assess economic development and define new policies for sustainable development. A model entitled “Economic Input-Output-Based Life Cycle Assessment, EIO-LCA” is to be developed with Portuguese economical and environmental data. EIO-LCA complements the economic input-output analysis by linking economic data with resource use (such as energy, ore and fertilizer consumption) and/or environmental output data (such as greenhouse gases emissions, toxic discharges, ozone depletion potential, hazardous or non-hazardous waste). The analysis will also includes quantification of the carbon balance of Eucalyptus globulus forests in typical regions of Portugal, including changes in stocks and fluxes during the whole rotation of the plantations.

d) Carbon Cycle

Correct perception of the global carbon cycle at the planetary scale is a major need in all environmental models. Additionally to other aspects mentioned elsewhere in this document, our LA can contribute, in a novel, multidisciplinary manner to this key issue. Our approach is to integrate seabed studies of carbon release/fixation in/off the oceanic crust and crustal production of methane and methane hydrates, with the dispersion of hydrothermal plumes (discrete and diffuse) generated under and on the seabed. This requires the development of analytical tools and procedures, and interpretative models [requiring the participation of the various units of the present LA, including Creminer and IN+]

e) Flows of materials in the economy from raw-materials extraction to final re-integration in the natural environment

Despite all efforts, it is more and more evident that unleaking final disposal facilities are impossible or extremely costly to build. This implies that a deep scientific knowledge of the processes involved in the final environmental dispersion of the generated chemicals is needed. The required studies cover the fields of structural geology, geophysics, hydrogeology and low-temperature inorganic and organic geochemistry. The aim of this multidisciplinary approach is to find the best geological conditions for disposal in terms of the physical and chemical properties of the substrate rocks and to choose the most economic and adequate monitoring methods for each case. Integrated with biological and toxicological studies available in the literature, this will enable a sound evaluation of the indirect environmental costs of this “post-life cycle” phase of manufactured goods.

f) Techno-economic assessment, technological change and systems integration

In the current socio-economic context, it appears that a main policy challenge is to promote innovation with competence building and social cohesion. At present various indicators are used to illustrate the structure and the changes of the science, technology and innovation systems and their impact on the economy and society. However, if any information can really be taken from the broad spectrum of values given for OECD countries, is that the design of science and technology policies must encompass the

30

careful analysis of the education systems, as well as the specific conditions following the trajectory of each country. This requires an increased accountability and observatory effort, which has clearly not been equally considered by every country, even those in OECD. We propose here to focus on the idea of innovation and its implication for a sustainable, globalizing learning economy.

Actions Proposed. Deliverables

ECO-auto: Ecodesign for Portuguese automotive components

Main deliverables include - 5 years: Development of a tool to support the design for disassembly of complex products, such as

automobiles and appliances. These products are generally composed of multiple components grouped in several subsets. This tool is aimed at optimising the product’s disassembly sequence, considering the following attributes for each product subset: precedence relations, dismantling costs (mainly associated to labour) and revenues associated with the disposal. An optimisation algorithm is to be developed to choose the disassembly sequence which maximizes environmental and economic criteria.

Mapping engine performances depending on vehicle use and loading, for a set of vehicles. Software enabling the evaluation of the Environmental Impacts of design innovations on car

components, over the Total Life Cycle.

The lack of resources initially planned to support the research work under this theme has affected the initial work plan as follows:

o Industrial Ecology Systems (including ECO-auto: Ecodesign for Portuguese automotive components)

Environment-economy interconnections were supposed to be detailed studied in terms of their relative dependence on economic activity fields or sectors, on existing local infrastructures and on future technological options, but this work has been affected by the lack of a post-doctoral researcher. In particular, the development of a software tool enabling the evaluation of the Environmental Impacts of design innovations on car components, over the Total Life Cycle, has been delayed due to lack of a specialized pos-doctoral researcher.

Low-Power Eco-Burning Equipment: design, monitoring and control of low-power burning systems

Main deliverables - 5 years: Fundamental analysis of the flame-flow-wall interaction for optimum development of “burner-

recipient” prototypes and the implementation of active control systems for eco-burning. Fundamental analysis of direct injection burning systems for the clean combustion of liquid fuels

in engines. Retrofitting of existing burners (low- to medium power) to use natural gas, low-calorific, and multi-

components blending fuels while burning under lean conditions. Development of a laboratory scale pulsed combustor with full control of inflow conditions and

tests with different type of solid fuels, simulating the waste hazardous compounds.

Main deliverables - 10 years: Optimized domestic burners, together with active control system Pulse incinerator system, as a function of fuel used and of relevant operating conditions. Detailed studies on the impact of the implementation of the techniques above: The adequacy of waste treatment techniques in the context of their input-output parameters and

cost/effectiveness factor. Adaptation/modification of thermal treatment systems as a short-time strategy for waste

reduction improvement. Suggestions on the potential use of more innovative solutions, based on modern actively

controlled systems regarding the combustions processes, as a long-time strategy. The impact of these “clean combustion systems” on the ratio of efficiency/cost taking into

account the highly demanding pollutant emission regulation.

31

The lack of resources initially planned to support the research work under this theme has affected the initial work plan as follows:

o Low-Power Eco-Burning Equipment: design, monitoring and control of low-power burning systems

The Laboratory work has been affected by the lack of a technician specialized on hardware developments (electronics).

ECOaccounting: EIO-LCA methodology for Kyoto accounting in Portugal

Main deliverables include - 5 years:

A software tool to promote EIO-LCA calculation. Development of a National Accounting Matrix for Economic Sector Environmental Burden

Characterisation. Analysis of the impact of different economic development scenarios on Global Warming

Contributions in Portugal, within the Kyoto framework.

Fostering the Kyoto forest in Portugal

Main deliverables include - 5 years:

Carbon balance of eucalypt plantations in Portugal, including:o Quantifying the net ecosystem carbon exchange through the continuous measurement of

surface flux of carbon dioxide using the eddy covariance method;o The quantification of carbon stocks by the inventory of biomass components and changes in

soil carbon storage along a chronosequence of eucalypt plantations;o To extrapolate the results found for carbon sequestration in the main site, across a range of

soil and climate conditions in Portugal.

Water use by eucalyptus forest ecosystem in Portugal, including: o Measurements of the water flux at different hierarchical levels: leaf (stomatal conductance),

tree (storm and boundary layer conductance), stand ( eddy fluxes of latent heat) and regional ( remote sensing);

o Development of an energy-water-carbon model to evaluate the Eucalyptus forest water use. This model will also be necessary to incorporate into the land surface schemes used in General Circulation Models, to properly assess the effect of this forest response on the global climate.

The lack of resources initially planned to support the research work under this theme has affected the initial work plan as follows:

o Industrial Ecology Toolbox (ECOaccounting: EIO-LCA methodology for Kyoto accounting in Portugal)

The establishment of a structured set of indicators to support sustainable policies and priority setting at a regional level is still at an early phase of development and was affected by lacking a post-doctoral researcher able to carry on with the detailed work. This has also affected the development of a new organization of infra-structures to promote co-operation between the various actors involved within an Industrial Ecology framework.

Oceanic Carbon Biogeochemical Cycle

Main deliverables include - 5 years:

32

Development of sampling, analytical and extrapolation tools for more precise evaluation of the oceanic carbon cycles related to seabed phenomena such as serpentinization and methane hydrate formation and release.

Main deliverables include - 10 years: Refined estimate of the oceanic carbon budget related to subseabed phenomena.

Flows of materials in the economy from raw-materials extraction to final re-integration in the natural environment

Main deliverables include: 5 years:

sampling protocols for soils, sediments and waters affected by waste disposal sites; software package for geochemical anomaly interpretation based on multifractal

techniques; chemical models for the adsorption of metals by geomedia;

10 years geological, geochemical and geophysical criteria for selection of disposal sites in

Portugal inland; recommended sites for (selective) disposal in Portugal inland; evaluation of fluxes and sinks of the main pollutent agents in different geological

settings in order to construct preliminary retention capability maps for Portugal inland;

Techno-economic assessment and prospective development of clusters

Main deliverables include - 5 years: Identification and assessment of technology platforms sustaining emerging clusters and

continuous monitoring of specific regional clusters in Portugal (e.g., Aveiro, Braga, Oeiras) and related training needs.

Industry-Science relationships: Continuous analysis of the institutional framework involved in the Portuguese catching-up process with emphasis on industry-science relationships, following main OECD guidelines.

Innovation and the environment: Continuous assessment of how environmental policies can stimulate technological innovation, and how innovation policies can enhance environmentally sustainable economic development.

Innovation and productivity: within the large field of possible issues associated with how innovation contributes to produce capital accumulation, we propose to focus on the relationship between innovation and productivity. In fact, productivity growth is, in the long run, the main driver of economic growth, and we suggest that it is important to complement generic research on how innovation contributes to productivity with specific analysis of technologies at the firm and sectorial level.

The lack of resources initially planned to support the research work under this theme has affected the initial work plan as follows:

o Technical Change and Systems of Innovation (including the techno-economic assessment and prospective development of clusters)

The work on the dynamics of the process of knowledge accumulation and techno-economic assessment was planned to consider the development and implementation of a new information system for the mapping of innovation and knowledge in Portugal (namely, in a European context) and, again, this part of the work plan was affected by the lack of a post-doctoral researcher. It should be noted that the main conceptual work has been concluded, but the systems has not been validated and implemented.

33

34

3.4.Thematic Area D: Signal Processing for Communication Networks and Multimedia

Signal, image and video processing techniques find applications in several areas of practical interest. The large technological development of the digital signal processing electronic systems has enabled the increase of their processing speed and of their integration capacity as well as the decrease of the costs. Since this tendency is expected to be even clearer in the future, the signal processing techniques appear as one of the fundamental tools to the development of the future communication systems. In particular, the kind of services supported by the future wireless network systems demand:

1. that they are adaptive in a way that optimises the utilization of the spectral resources (bandwidth);2. advanced signal processing techniques (to achieve that goal), in particular the development of

(i) advanced algorithms for compression of video and voice,(ii) advanced algorithms for power control, channel estimation, interference cancellation, synchronization, coding/decoding, etc.,(iii) smart antennas,(iv) development and analysis of complex protocols for network traffic management,(v) user-friendly human-machine graphical interfaces,(vi) speech recognition,(vii) etc.

3. and that these algorithms run in heterogeneous configurable digital signal processing architectures.

The driving forces whose evolution is illustrated in Figure 1 condition the use of advanced signal processing methods1. The growing rate of the algorithmic complexity, as imposed by the several generations of communication systems, is higher than the one of the processors performance (processors speed). Thus, independently of the future evolution of the processing architectures, the degree of sophistication of the signal processing methodologies make clear the extreme relevance of optimising the complexity of the resulting algorithms. These goals are particularly important when designing and producing dedicated architectures for communications and multimedia signal processing.Communication networks, in particular the Internet, require automatic tools for traffic estimation, characterization, and prediction. These are signal processing applications supported in stochastic models. It is important to identify the influence of specific traffic patterns on the global performance. This identification conditions investment decisions for infrastructures that support that traffic. Also, Internet related research teams need traffic characterization tools to simulate real scenarios in test situations. This way, whoever may be managing and planning data network investments will have rigorous and high quality information to guide the decisions.

1 From: Heinrich Meyr, Form follows Function: Mapping Algorithms to Architecture, Plenary Session, 2001 IEEE Third Workshop on SIGNAL PROCESSING ADVANCES IN WIRELESS COMMUNICATIONS.

35

1

10

100

1000

10000

100000

1000000

10000000

1980 1988 1996 2004 2012 2020

Algorithmic Complexity(Shannon's Law)

Processor Performance(Moore's Law)

Batery CapacityG1

G2

G3

Figure 1: Algorithmic driving force

All the actions herein proposed, either for immediate execution or for preliminary exploration, have strong links with the other thematic areas. In particular, both the thematic areas A and B plan to work with cooperating robots that constitute a robot network and requires advanced communication techniques based on efficient signal processing methodologies. Also, the automatic construction of 3D visual models is relevant to the performance of the recognition and navigation systems.

Proposed Actions. Deliverables.All the actions in this thematic Area D share a common guideline whose goal is to develop advanced signal processing algorithms with low computational complexity. To achieve this goal, we use new methodologies based on adequate mathematical tools such as the analysis on the time-frequency plane, optimization theory, non-linear algebra, and differential geometry.

Action 1 - Signal Processing for Wireless Communications Radio SystemsThe services of the future generations of wireless communications systems require a significant increase of capacity in terms of the number of simultaneous users sharing wider spectral bands. The following topics, being very actual and with high development potential in the years to come, give a decisive contribution to the significant increase of the system capacity.

a)Blind source separation

(i) We approached the blind source separation problem within the context of spatial division multiple access (SDMA) networks for wireless networks. In SDMA architectures, several users within the same geographic cell share the same time–frequency channel. This efficient spectral allocation strategy per cell permits to expand the overall capacity of current cellular infrastructures, without consuming additional radio frequency (RF) bandwidth. From the receiver viewpoint, the SDMA technique raises a new signal processing problem: the SDMA receiver has to separate the linearly superimposed users. This is in sharp contrast with more traditional multiple-access schemes such as TDMA or FDMA, where the users do not crosstalk due to their explicit separation in either the time or the frequency axis. More importantly, in order to not waste precious bandwidth resources in sending training sequences, an efficient SDMA receiver must have the capability to isolate the signal from each user in a blind manner, that is, without knowing the space-time linear matrix channel which mixes the emitted signals before reception. In other words, the SDMA receiver faces a classical blind source separation problem. In our research, we exploited the 2nd order statistics (SOS) of the observed data and the prior probability density function (pdf) associated to the unknown channel matrix, that is, we work under a Bayesian framework. It is well known that, in a deterministic framework (channel matrix is unknown but deterministic), the SOS of the observed data permit to solve for the channel matrix up to a right unitary matrix. Under a Bayesian framework, this residual unitary matrix is now a random object. Its associated pdf depends solely on the prior pdf on the channel matrix and is easily computable, resorting to some differential-geometric based techniques in Random Matrix Theory. Our work capitalizes on this statistical knowledge: we incorporate the pdf of the residual unitary matrix into the locally convergent algorithm, which implements the joint (maximum a posterior) MAP channel/source estimator. Our computer simulations show that a better global convergence rate is obtained when compared to traditional schemes. To implement our estimator we developed geodesic descent optimization algorithms to quickly locate the missing unitary channel factor within Lie group of unitary matrices.

(ii) If the active wireless users within a network cell possess high mobility, then the time-invariance assumption of their space-time signature (channel) is no longer realistic and must be tracked (estimated) in parallel with the detection of the information symbols. We examined this problem, that is, the joint source symbol detection and multi-channel acquisition problem, in the context of wireless digital flat-fading links with space diversity (as in SDMA architectures). To begin and gain intuition, we only have studied, at this point, the single-user scenario. The originality of our approach relies on a statistical model which decouples the time dynamics of the multi-channel vector in amplitude and direction. The

36

time evolution of the direction of the channel vector is a 1st-order Markov process on the unit-sphere, with transition kernel governed by a von Mises distribution. We compute the most probable emitted symbol sequence and channel realization for this statistical model, given the set of availabe array observations. Our maximum a posterior (MAP) receiver consists of a bank of parallel processors. Each processor finds the most probable channel realization for a given symbol sequence via a differental-geometric second order geodesic descent scheme that fully exploits the curvature of the constraint surface (in our case, a multi-dimensional torus, a curved manifold in which a channel trajectory is a point). Our preliminary results assess that our channel model can approximate the statistical behaviour of certain well-known wireless channels (e.g. Jake’s model, Rice model, etc) even in the presence of Doppler spread. Also, computer simulations confirm that the differential-geometric based methods that we developed outperform (in terms of accuracy versus complexity) the traditional optimization schemes (e.g., constrained Newton method).

Deliverable 1.(i)Continued production, during the contract time, of new advanced algorithms to blind source separation. These algorithms are based on innovative technologies and optimized in what respects to the computational complexity.

b) Space/time coding

We addressed the problem of space-time communication from the most difficult perspective: nonchoerent communication. That is, channel state information is absent at both ends of the communication link (the transmitter and the receiver). A possible structure for nonchoerent receivers (although not mandatory) is a first processing step in which channel identification is attempted; this channel measurement is then exploited to decode the information symbols contained in the received data packet. Restricting ourselves to channel estimation based on 2nd order statistics (SOS), it turns out that the matrix channel is only identifiable up to phase shift in each column. This ambiguity complicates the study of the performance analysis of any given proposed SOS-based channel estimator, as the traditional bounding tool – the Cramer-Rao Bound (CRB) – becomes theoretically infeasible. In fact, it is a recurrently observed fact that many parametric statistical models suffer from ``intrinsic ambiguities" in the sense that the distribution of the observation vector is invariant to smooth, structured changes in the model's parameters. A typical situation occurs in overparameterized statistical models. The fact that certain members of the parametric statistical family are locally undistinguishable makes the Fisher information matrix (FIM) associated to the given statistical model singular. We examined such degenerate deterministic parameter estimation problems from a Riemannian geometric perspective. Our idea consists in replacing the original (ambiguous) parameter set by a lower-dimensional Riemannian (non-ambiguous) parameter set. The new parameter set comes in the form of a quotient space and is obtained by identifying equivalent family members in the initial parameterization. We developed extensions of the Cramer-Rao bound (CRB) for this Riemannian setup. As a byproduct, fundamental limits for the intrinsic variance of unbiased estimators in parametric estimation problems with a priori deterministic constraints were obtained. Our new results take into account the traditional Fisher information of the observation data, but also the curvature of the Riemannian manifold which acts as the parameter space.

Deliverable 1.(ii) In 5 years: preliminary results report. In 10 years: software for space/time coding algorithms.

c) Future work The Riemannian viewpoint in the context of nonchoerent communication seems to be very promising, in light of our preliminary results on performance analysis. We intend to pursue this line of work and extend the ideas of Riemannian quotient spaces to the problem of

37

specific space-time code design. This involves also generalizing some well-known concepts and facts from Information Theory in Euclidean spaces to this new mathematical setting.

Action 2 - Underwater Acoustic Communications SystemsThe growing interest in ocean exploration and the recent development of autonomous surface vehicles and submarines that perform inspection tasks in cooperation require efficient and robust data communication systems. Digital communication using acoustic modems is the method of choice for exchanging data among distant or highly mobile equipment used in various underwater activities. However, achieving efficient communication in this environment is challenging due to severe distortions that affect the transmitted signals as they undergo multiple reflections and refractions in their propagation path. Attempts to overcome these impairments in high data rate coherent modems haven't been entirely satisfactory, thus hampering their widespread adoption. Research on robust underwater communication systems followed two main research directions, namely, (i) adaptation of state-of.the-art multicarrier techniques developed for wireless radio systems, and (ii) exploitation of intrinsic properties of acoustic wave propagation in the ocean to mitigate the effects of multipath or improve the robustness of signal detection and demodulation.

a)Multicarrier communicationsRecently, much attention has been devoted to the use of OFDM (Orthogonal Frequency Division Multiplexing) modulation for wireless and cable applications as a way of approaching channel capacity with simple transmitter/receiver architectures. In OFDM the message stream is divided into many parallel lower rate streams that modulate a set of partially overlapping orthogonal carriers. Since longer symbols are less sensitive to multipath, equalization requirements may be considerably relaxed on each subcarrier. This feature is quite appealing in underwater communications, where highly complex filters used for equalization constitute the main computational bottleneck. Motivated by research on space-time modulation/coding in wireless radio channels using multiple transmit and receive antennas, the MIMO (multiple input/multiple output) framework was adopted as a means of obtaining new equalizer structures. Through this rich framework, OFDM transmission can be tackled in scenarios where multiple projectors and hydrophones are used at the transmitter and receiver, respectively. When compared with other approaches, the proposed receiver structure provides a direct extension of known single-carrier/single-user results based on decision-feedback equalization, thus retaining the ability to handle severe multipath distortion.

b)Physics-based signal processing

Time-reversed focusing by using multiple transmit/receive elements was proposed as a means of reducing the effects of intersymbol interference in digital communication signals, including OFDM-modulated waveforms. This feedback process takes advantage of the reciprocity of the propagation medium, and achieves transparent multipath compensation without detailed knowledge of the environment. Its effectiveness was demonstrated in single-carrier communications, even in the presence of time variations due to receiver motion. The technique can be readily applied to multicarrier communications as well, where it enables cyclic-prefix-based demodulation with modest time redundancy. When the number of available transducers is insufficient to properly focus the desired OFDM waveform, MIMO equalization was shown to be a suitable receiver structure. This approach effectively distributes the signal processing load between the transmitter and receiver. Research on signal processing techniques that take into account the propagation environment also addressed passive detection and classification of underwater acoustic transient signals. The focus is on theoretical and algorithmic aspects so as to achieve an acceptable compromise between optimality and robustness to model mismatches, and computational efficiency. Several classes of transients are considered, e.g., man made, mammals' signatures, and spiky noise generated by hydrothermal vents. The insights gained from this work could form the basis of optimal/efficient receiver architectures that take into account the random nature of some ocean processes that distort digital communication signals. Previous work showed that local stationarity was important to achieve the efficient detection of transient processes. When the processes are locally stationary, the wavelet transform is efficient for real-time detection. The results obtained

38

showed that the existence (or not) of local stationarity could be observed either in the Time-frequency plane (or Wigner distribution) or the two-dimensional power spectrum, both derived from the process two-dimensional autocorrelation function. Research then focused on obtaining a simple algorithm that would allow for the estimation of locally stationary autocorrelation functions (or covariance matrices in the discrete-time domain) from data. It was shown that although, in general, the estimates of covariance matrices that concentrate the maximum of the signal energy in the smallest number of nonzero eigenvalues are not locally stationary, it is always possible to obtain a stationary description of the process to detect. An easy-to-implement algorithm to obtain a locally stationary covariance matrix from a non-locally stationary one was derived, and the corresponding efficiency gains were assessed.

c) Software-defined radio

As part of this research effort, a prototype underwater acoustic digital communication system is to be developed. The system will be based on the software-defined-radio approach, whereby digital signal processing is used not only on the base band but also as high as possible in the frequency conversion chain. This enables a much higher flexibility and adaptability of both the emitters and the receivers and the accommodation of several modulation schemes, equalization and synchronization techniques, and codification methods.The survey of available building blocks has been completed. Whenever possible, low-cost commercially-available equipment has been chosen. The system will be designed around a low-end floating-point DSP board from Texas Instruments. A Signalware daughtercard will provide A/D and D/A conversion, while conversion between (digital) base-band and pass-band will be performed by another daughtercard developed in-house. The latter will be based on upconverter and downconverter ASICS from Analog Devices. The design of prototype analog electronics will be kept to a minimum, mainly for signal conditioning purposes; projectors, hydrophones, and their associated drive electronics will be purchased from a transducer manufacturer. Mechanical encapsulation will draw from the expertise developed by other research labs at the Institute for Systems and Robotics. The basic software functionality for a real-time single-carrier modem has been completed. This includes packet generation (framing), error-correction coding, equalization and timing and carrier recovery.

Deliverable 2. In 5 years: first version of a prototype of an underwater acoustic digital communication

system, based on the software radio paradigm, using advanced and efficient signal processing algorithms, and implemented in dedicated hardware.

In 10 years: upgrade of the prototype first version, using efficient methodologies adapted from the modern radio communication systems, with flexibility to accommodate several modulation schemes, equalization and synchronization techniques, and coding methods.

a)Future Work

Regarding MIMO equalization algorithms for OFDM transmission, future work will seek to develop coefficient adaptation algorithms, characterizing the performance of the receiver in terms of residual error, tracking speed, and robustness to numerical errors. So far, the effect of Doppler shifts induced in acoustic waveforms even by relatively slow emitter/transmitter motion caused by waves and currents has not been studied in detail, but will likely play an important role in underwater OFDM systems. Future work in this area will seek to develop differential Doppler estimators, and integrate them into MIMO equalizer structures using delay-Doppler spread functions, similarly to what has been accomplished by other authors for single-carrier transmissions. The study and mitigation of the degradation in performance caused by environment fluctuations and motion on OFDM waveforms is also one of the relevant research topics in the context of time-reversed focusing.Regarding the detection of transient signals, the tests performed with real data showed that using a local stationary second-order model of a nonstationary process was useful for the detection of transients in noise, when the signal to detect had either a small or large number of relevant eigenvalues. In fact, it was shown that for real-time detection of transients, or when the time localization of the transients was unknown, the use of a locally

39

stationary estimate of the covariance matrix of the signal to detect lead to computationally efficient detectors, at only a small performance degradation cost.The results obtained with one-dimensional data suggest that such results can be extended to the 2-D case. In many applications such as remote sensing, mobile robotics or automatic surveillance, the detection of features in images or image sequences is typically a computationally costly procedure. Therefore, we propose to adapt and apply the local stationarity results for 1-D to the 2-D situation, and study the problems related to feature modelling and robustness to noise and geometrical modelling errors. The results must compare the proposed methods with near-optimal but computationally heavy classical procedures in order to clearly assess the computational gains and performance degradation. Regarding the development of a prototype modem, some systems integration work needs to be done to assemble the core computational engine comprising A/D and D/A conversion, up- and downconversion, and DSP processing. This entails the development of some digital hardware, although an effort was made to keep this at a minimum by resorting to programmable logic (FPGA’s) whenever appropriate. A market survey should be conducted to locate suppliers for the required transducers and associated electronics, which have already been fully specified. Analog electronics for automatic gain control and filtering should be developed as well, although the task is greatly simplified by the high sampling rates tolerated by the baseband/passband converter ASICS, which exceed the desired frequency band by several orders of magnitude. Regarding software, the currently available single-carrier solution should be upgraded to handle multiple data streams, as found in OFDM waveforms. This task will only be carried out after a working single-carrier prototype has been demonstrated.

Action 3 - Acquisition of 3D Models of Real Environments

a)Automatic 3D modeling from video

The successful completion of the objectives refered in this topic hinge on the development of reliable image based tracking and matching processes, together with observer motion estimation capabilities. Steps towards this objective were given through developments at the level of image to image matching on one side, and motion estimation without correspondences on the other side. The matching techniques result from relaxating the original combinatorial optimization problems into concave minimization problems which can be solved with much more efficient techniques. Though highly efficient relative to the original problem, the resulting nonlinear problems are still practically unfeasible as we scale up the number of points envolved in the matching process. We all know that images can easily have tenths of thousand of (matching) points, leading to unsurmountable complexity. Future Work.The strategy, and the resources which lacked so far, must be directed into searching for reasonable simplifications, leading to linearized/linear problems which can be solved efficiently, even with sizes like those mentioned before. A great effort must be done in modelling the 3D structure estimation process, so that both prior knowledge and data from several sources can be used. Specifically, previous reconstructions (approximate reconstructions of the scene) and combination between geometric and radiometric information (ex. Rigidity and image texture) must be included in the global matching process.One clever way of reducing complexity is assuming known simple facts about the world structure. For example: “it’s man-made therefore rich in planes and lines”! We have shown that, assuming the existance of lines in the 3D world, it is possible to compute observer motion (orientation) without resorting to high-complexity matching. Further investigation is required on how to recover in a similar fashion other motion parameters.

b) CODEC using 3D representations: data generation, low bit rate transmittion, and reconstruction

Our developments towards de goal of coding video using 3D representations rely, essentially, on the possiblity of rendering one image (projection), from knowledge of the point of view (position and orientation) of the camera and a 3D model of the object. Several problems arise in this context, namely the incompleteness of data: partial views of the object, missing information, occlusion and inclusion. Recently we have shown how to handle most of these problems and by now we are able to generate full 3D models and

40

render “virtual” images, even with parts of the object missing from the video sequence. The framework hinges on iterative algorithms which estimate missing data by using knowledge about the imaging process. They have proved reliable in general situations.Future Work. Provided the necessary resources are put into this problem, further developments are necessary towards the final goal of generating more efficient 3D scene models. Among others, some work must be devoted to the problem of associating non consecutive views of the same object which, in current methods, are considered separate objects. Also, convergence properties of these algorithms, which may limit its applicability, have to be studied.

Deliverable 3. In 5 years: system for automatic 3D modeling from video. This system is relevant, for

example, to the automatic virtualization of architecture.

Deliverable 4. In 5 years: CODEC, using 3D representations, to the generation, low bit rate

transmittion, and reconstruction, based on algorithms with optimized computational complexity and possibly running in dedicated hardware. In this phase, we consider an observer moving through a static scene.

In 10 years: evolution of the system to cope with general scenes with multiple moving objects. This scenario is relevant to the analysis of video sequences of underwater environments.

Action 4 -Statistical Analysis of Traffic in the Internet

This section describes the work plan whose goal was at the beginning to potentate the creation of a new action. Although we had enough resources and critical mass, we have identified the need to have a preliminary exploratory phase for this area of intervention so that, according to the results of this preliminary phase, we could decide about the definite proposal of a new action in the area of the Statistical Analysis of the Internet Traffic. Therefore, we have started the development of this area in 2002. Preliminary work was developed in the topic of Queuing Networks. In particular, we focused our research on determining a probably stabilizing regulator for distributed scheduling policies for multi-class queuing networks. Along 2003, we investigated whether that regulator could improve performance over existing policies when stability is not a major concern.Our own evaluation of the work developed along 2002 and 2003 shows that the activity in that area was not so promising as we could expect 4 years ago. Although it was an area under initial development, the fact is that, when compared with the other 3 actions, the results achieved are far from being satisfactory. As so, the coordinator of the thematic area D is equating to close the research activity in the area of the Statistical Analysis of the Internet Traffic. This should not been interpreted as a final and definite decision. If the necessary conditions, in terms of post-docs contracts and graduate students grants, can be guaranteed, that action can re-activated in the future.

41