2005 Technology and Persons with Disabilities conference Assistance Robotics Implementation of...

Post on 05-Jan-2016

213 views 0 download

Tags:

Transcript of 2005 Technology and Persons with Disabilities conference Assistance Robotics Implementation of...

2005 Technology and Persons with Disabilities conference

Assistance Robotics

Implementation of human-like visuo-motor synergies on a teleoperated mobile device

Daniel R Mestre1, Yves Rybarczyk1, Philippe Hoppenot2, Etienne Colle2

 1Mouvement & Perception Lab (LMP), CNRS, Marseille,

France

2Complex Systems Lab (LSC), CNRS, Evry, France

E-mail: mestre@laps.univ-mrs.fr

LMP

2005 Technology and Persons with Disabilities conference

ARPH : Assistance by Robot for People with Handicap, in collaboration with AFM

(Association Française contre les Myopathies)

2005 Technology and Persons with Disabilities conference

Command control station

ARM: Assistive Robotic Manipulator

Pan-tilt camera

Ultrasonic ring

Odometer

2005 Technology and Persons with Disabilities conference

Exact Dynamics b.v.

(www.exactdynamics.nl)

Booth 203, Marriott

2005 Technology and Persons with Disabilities conference

Command control station

Assistive Robotic Manipulator

Pan-tilt camera

Ultrasonic ring

Odometer

2005 Technology and Persons with Disabilities conference

Control station terminal

2005 Technology and Persons with Disabilities conference

Global scheme of operation in autonomous mode

Goal-seeking behavior

Avoidance behavior

2005 Technology and Persons with Disabilities conference

Khepera

Planned pathOptimal path computed

Actual path

2005 Technology and Persons with Disabilities conference

Advantages of the approach

Cost-effective

Low quality captors

Human-like (compatible)

Hierarchical strategy (planning & navigation)

Avoidance and goal-seeking behaviors

Linear speed is inversely related to angular speed

2005 Technology and Persons with Disabilities conference

Viviani & Terzuolo (Neuroscience, 1982)

Movement’s velocity is proportional to the radius of curvature

2005 Technology and Persons with Disabilities conference

Disadvantages of the autonomous approach

Errors due to the poor quality of captors

Full automation is not possible

Exploration, changes in plan during action,…

2005 Technology and Persons with Disabilities conference

Human-machine cooperation is needed

Task allocation

Appropriation (distal attribution)

Human like behaviors

1. Speed/curvature relationships2. Gaze/motion relationships (on the robot)3. Camera control (human interface)

2005 Technology and Persons with Disabilities conference

Land, M.F. & Lee, D.N. (1994)

Where we look when we steer

2005 Technology and Persons with Disabilities conference

2005 Technology and Persons with Disabilities conference

J. Wann (2004)

2005 Technology and Persons with Disabilities conference

Distribution of gaze location on the road

2005 Technology and Persons with Disabilities conference

Experimental study

2005 Technology and Persons with Disabilities conference

Task : slalom w/ different curvatures

2005 Technology and Persons with Disabilities conference

Condition 1: fixed camera

Decrease linear speed

Increase linear speed

Increase left angular speed

Increase right angular speed

2005 Technology and Persons with Disabilities conference

Robot’s mode of control (fixed camera)

Angular speed

Linear speed

Operator’s commands

trajectory

2005 Technology and Persons with Disabilities conference

Condition 2: mobile camera

Decrease linear speed

Increase linear speed

Cameraleft

Camera right

2005 Technology and Persons with Disabilities conference

Robot’s mode of control (mobile camera)

Camera angle (CA)

Angular speed (AS) AS= f(CA)

Linear speed (L) L = 1 - |AS|

Operator’s commands

trajectory

2005 Technology and Persons with Disabilities conference

Results

2 groups of 7 subjects (healthy volunteers)

After training, 4 trials on the slalom

2005 Technology and Persons with Disabilities conference

Sessions

1 2 3 4

Com

ple

tion

tim

e (s

ec)

0

50

100

150

200

250

300

350fixed cameramobile camera

Required time to complete the task

2005 Technology and Persons with Disabilities conference

Fixed camera condition

Trajectories are jerky

Sharp curvesStraight lines

2005 Technology and Persons with Disabilities conference

Mobile camera condition

Trajectories are smooth

2005 Technology and Persons with Disabilities conference

Radius of curvature distribution

Radius of curvature (m)

0.0001 0.001 0.01 0.1 1 10 100 1000

Occ

urre

nce

(%)

-10

0

10

20

30

40

50

60

70mobile camerafixed camera

2005 Technology and Persons with Disabilities conference

fixed camera

radius of curvature (m)

0.1 1 10

Lin

ear

spe

ed

(m/s

ec)

0.2

0.3

2

1

2005 Technology and Persons with Disabilities conference

Time (tics)

0 100 200 300 400 500 600

0.1

1

10

radius (m)

Speed (m/sec)

2005 Technology and Persons with Disabilities conference

Mobile camera

Radius of curvature (m)

0.1 1 10

Lin

ear

spe

ed (

m/s

ec)

0.5

1.5

2

1

2005 Technology and Persons with Disabilities conference

Time (tics)

0 100 200 300 400 500 600

0.1

1

10Radius (m)Speed (m/sec)

2005 Technology and Persons with Disabilities conference

Speed=1.3Radius1/3

Radius of curvature (m)

0 2 4 6 8 10

Line

ar S

peed

(m

/sec

)

0

1

2

3

4

We observe the « power-law » described by Viviani et al. In natural movements

2005 Technology and Persons with Disabilities conference

This type of study has to be pursued, notably because

we tested here only "healthy" young volunteers

This methodology has to be adapted

to handicapped populations

Conclusions (1/3)

2005 Technology and Persons with Disabilities conference

However, this experimental approach suggests

1) that the implementation of "human-like" visuo-motor coupling models on complex assistance devices

might improve human-machine compatibility

Conclusions (2/3)

2005 Technology and Persons with Disabilities conference

Further, this experimental approach suggests

2) that, beyond raw performance,

the evaluation of the similarity between teleoperated

movements and "natural" (non-mediated) movements

might constitute an interesting way to foresee the usability

of an assistance device

Conclusions (3/3)