Feedback Systems in Engineering and Nature · uncertainty in complex systems ... •Many molecular...
Transcript of Feedback Systems in Engineering and Nature · uncertainty in complex systems ... •Many molecular...
Feedback Systems in Engineering and Nature
Richard M. MurrayControl and Dynamical Systems
California Institute of Technology
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Outline
I. Feedback and Control TheoryII. Applications in Science & EngineeringIII. Linkages to Computer Science
(Grand Challenges)IV. Summary
Jean Carlson (UCSB) John Doyle (Caltech) Naomi Leonard (Princeton) Simon Levin (Princeton)
Jerry Marsden (Caltech) Linda Petzold (UCSB)
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 3
What is Feedback & Why is it Important?Feedback = mutual interconnection of two (or more) systems• System 1 affects system 2• System 2 affects system 1• Cause and effect is tricky; systems
are mutually dependent
Feedback is ubiquitous in natural and engineered systems• Major mechanism for managing
uncertainty in complex systems• Often leads to “robust yet fragile”
behavior• Claim: Understanding complex
systems requires an understanding of feedback
Terminology
System 2
System 1
System 2System 1
System 2System 1
ClosedLoop
OpenLoop
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 4
Example #1: Flyball Governor“Flyball” Governor (1788)• Regulate speed of steam engine
• Reduce effects of variations in load (disturbance rejection)
• Major advance of industrial revolution
Balls fly out as speed increases,
Valve closes,slowing engine
http://www.heeg.de/~roland/SteamEngine.htmlBoulton-Watt steam engine
Flyballgovernor
Steamengine
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 5
Two Main Principles of FeedbackRobustness to Uncertainty through Feedback• Feedback allows high performance in the
presence of uncertainty• Example: repeatable performance of
amplifiers with 5X component variation• Key idea: accurate sensing to compare
actual to desired, correction through computation and actuation
Design of Dynamics through Feedback• Feedback allows the dynamics (behavior)
of a system to be modified• Example: stability augmentation for highly
agile, unstable aircraft• Key idea: interconnection gives closed
loop that modifies natural behavior
Control involves (computable) tradeoffs between robustness and performance X-29 experimental aircraft
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Flight behavior in Drosophila(a) Cartoon of the adult fruit fly showing the
three major sensor strictures used in flight: eyes, antennae, and halteres (detect angular rotations)
(b) Example flight trajectories over a 1 meter circular arena, w/ & w/out internal targets.
(c) Schematic control model of flight system
6
Example #2: Insect Flight
More information:• M. D. Dickinson, Solving the mystery of insect
flight, Scientific American, June 2001
Different architecture than engineering• Large collection of diverse sensors
(many more than required)• Very slow computation with lots of
parallel pathways
Card and Dickinson (Caltech)
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 7
Feedback in Nature and Society
Biological Systems• Physiological regulation (homeostasis)• Bio-molecular regulatory networks
Environmental Systems• Microbial ecosystems• Global carbon cycle
Financial Systems• Markets and exchanges• Supply and service chains ESE
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 8
Control Theory: 1940-2000Modeling• Input/output representations for
subsystems + interconnection rules• System identification theory and
algorithms • Theory and algorithms for reduced
order modeling + model reduction
Analysis• Stability of feedback systems,
including robustness “margins”• Performance of input/output systems
(disturbance rejection, robustness)
Synthesis• Constructive tools for design of
feedback systems• Constructive tools for signal
processing and estimation
Basic feedback loop
• Plant, P = process being regulated• Reference, r = external input (often
encodes the desired setpoint)• Disturbances, d = external environment• Error, e = reference - actual• Input, u = actuation command• Feedback, C = closed loop correction• Uncertainty: plant dynamics, sensor
noise, environmental disturbances
r C(s) ++
d
ye
P(s)u
-1
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 9
Canonical Feedback Example: PID Control
Three term controller • Present: feedback proportional to current
error
• Past: feedback proportional to integral of past error
- Insures that error eventual goes to 0- Automatically adjusts setpoint of input
• Future: derivative of the error- Anticipate where we are going
PID design• Choose gains k, ki, kd to obtain the
desired behavior• Stability: solutions of the closed loop
dynamics should converge to eq pt• Performance: output of system, y,
should track reference• Robustness: stability & performance
properties should hold in face of disturbances and plant uncertainty
r C(s) ++
d
ye
P(s)u
-1
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Example #3: Chemotaxis
Chemotactic response has “perfect adaptation”• When exposed to a high attractant level,
receptors are methylated to provider higher ligand binding
• Result: activity level returns to nominal level in presence of a constant input
• Form of integral feedback (Yi et al, 2004)
10
http://www.genomics.princeton.edu/ryulab
Rao, Kirby and ArkinPLoS Biology, 2004
Sensing
Actuation
Computation
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 11
Robust Yet Fragile BehaviorFeedback shifts uncertainty from process to sensors• Provide robust behavior using accurate
sensors combined with feedback• Example: 10x uncertainty in transistors, 2%
uncertainty in resistors
Feedback provides robustness to some types of uncertainty, fragility to others• Fundamental tradeoffs between robustness,
performance, and stability• Higher performance in certain pathways
leads to increased sensitivity in other pathways
• Can be made rigorous through conservation laws in control theory
• Example: biological systems resistant to known environmental effects, but susceptible to new diseases
Control
Δ
Sensor
Process
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 12
Future Directions: Control in an Information Rich World1. Executive Summary2. Overview of the Field
• What is Control?• Control System Examples• Increasing Role of Information-
Based Systems• Opportunities and Challenges
3. Applications, Opportunities & Challenges• Aerospace and Transportation• Information and Networks• Robotics and Intelligent Machines• Biology and Medicine• Materials and Processing• Other Applications
4. Education and Outreach5. Recommendations
SIAM, 2003
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 13
Information and NetworksPervasive, ubiquitous, convergent networking• Heterogeneous networks merging communica-
tions, computing, transportation, finance, utili-ties, manufacturing, health, entertainment, ...
• Robustness/reliability are dominant challenges• Need “unified field theory” of communications,
computing, and control
Many applications• Congestion control on the Internet• Power and transportation systems• Financial and economic systems• Quantum networks and computation• Biological regulatory networks and evolution• Ecosystems and global change
Control of the network
Control over the network
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 14
Biology and Medicine“Systems Biology”• Many molecular mechanisms for biological organisms are
characterized• Missing piece: understanding of how network
interconnection creates robust behavior from uncertain components in an uncertain environment
• Transition from organisms as genes, to organisms as networks of integrated chemical, electrical, fluid, and structural elements
Key features of biological systems• Integrated control, communications, computing• Reconfigurable, distributed control, at molecular level
Design and analysis of biological systems• Apply engineering principles to biological systems• Systems level analysis is required• Processing and flow of information is key
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 15
Ecological SystemsWhat makes ecosystems robust?
Understand well enough to help manage ecosystems?• System evolution emerges from the interplay of processes at
diverse scales
• Explain robustness in terms of evolutionary forces acting at levels well below that of the patterns we observe
Case Study: Marine Ecosystems and Fisheries Management• Fisheries have not been robust to harvesting
• Fish feed on zooplankton and phytoplankton
• Plankton are patchy on almost every scale
• Understand spatial patterns, dynamics, robustness
Case Study: Wildfire management• Derive physical fire spread constitutive laws
• E.g: Fire/atmosphere coupling, dynamics, and feedback
• Wild land management, policy, training, and hazard mitigation
Some Themes: Role of Multi-Scale Dynamics and Feedback
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 16
CDS Panel Recommendations1. Substantially increase research aimed at the integration of
control, computer science, communications, and networking.
2. Substantially increase research in control at higher levels of decision making, moving toward enterprise level systems.
3. Explore high-risk, long-range applications of control to areas such as nanotechnology, quantum mechanics, electromagnetics, biology, and environmental science.
4. Maintain support for theory and interaction with mathematics, broadly interpreted.
5. Invest in new approaches to education and outreach for the dissemination of control concepts and tools to non-traditional audiences.
http://www.cds.caltech.edu/~murray/cdspanel
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 17
Grand Challenge #1: Autonomy and RoboticsAutonomous Desert Race• 175 miles; trails, dirt roads, open terrain• Vehicle must be completely autonomous• Corridor specified 2 hours in advance;
defines boundaries of race• First vehicle to complete the course in less
than 10 hours wins $2M
2007: Urban driving• Lanes, intersections, parking lots, other cars
Major challenges• Perception – what are we looking at• Decision making – where should we go• Technical driving – rough roads, slipping• System integration – getting everything to
work together
Future directions: service robotics, human/robot interaction
Barstow
Primm
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Grand Challenge #2: Control Using Slow ComputingTask: find food• Take off, avoid obstacles, find plume,
track plume to source
Approach #1: traditional control• Trajectory generation + feedback
• Requires substantial sensing and computing (ala Alice)
Approach #2: sensory-motor cascade• 300-500k neurons with
millisecond response times
• Large number of redundant sensors
• Excellent robustness, very good performance (over time)
18
approach &landing
take-off
visualnavigation
odortracking target
recognition
collisionavoidance
local search&
exploration
Richard M. Murray, Caltech CDSNCWIT, 14 May 08 19
Grand Challenge #3: Synthetic Biology
MIT Bio-Bricks program
Crawling Neutrophil “Chasing” a Bacterium • Human polymorphonuclear leukocyte (neutrophil) on blood film• Red blood cells are dark in color, principally spherical shape. • Neutrophil is "chasing" Staphylococcus aureus micro-organisms, added to
film. Tom Stossel, June 22, 1999 http://expmed.bwh.harvard.edu/projects/motility/neutrophil.html
Synchronization of a repressilator, IAP ‘03
7 Jan
14 Jan
Elowitz and Lieber, 2000
21 Jan
28 Jan
Elowitz and Lieber, 2000
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Mathematical Modelswith Uncertainty
Emerging Challenge: ModularityObservation: biological circuits do not compose well
20
Del Vecchio et al,Mol Sys Bio, 2008
Richard M. Murray, Caltech CDSNCWIT, 14 May 08
Feedback Systems in Engineering and NatureEmerging challenges in analysis and design of complex systems (natural and engineered)
• Develop the mathematical framework for unraveling the organization of highly constrained, uncertain, large scale, nonlinear networked systems
• New design tools for managing complexity, uncertainty, scalability, performance
Past successes of computer science are needed in new areas• Techniques for verification of large-scale, complex systems• Modular design methods that allow abstraction, composition
Further Reading: • Åström and Murray, Feedback Systems: An Introduction for
Scientists and Engineers. Princeton University Press, 2008. - Electronic: http://www.cds.caltech.edu/~murray/amwiki
• U. Alon, An Introduction to Systems Biology. Chapman & Hall/CRC, 2007
• M. D. Dickinson, Solving the mystery of insect flight, Scientific American, June 2001