Dimitrios Hristu-Varsakelis Mechanical Engineering and
Institute for Systems Research University of Maryland, College Park
http://glue.umd.edu/~hristu [email protected] Joint work with: M.
Egerstedt, S. B. Andersson, C. Shao. P.R. Kumar, P. S.
Krishnaprasad, Cooperative Optimization and Navigation
Problems
Slide 2
Outline Ensembles of autonomous vehicles operating on expansive
terrain. Bio-inspired trajectory optimization Language-based
navigation Report on Progress Event-driven communication
Slide 3
Slide 4
Examples from biology (bees, ants, fish etc.) Ensembles can
accomplish tasks that are impossible for an individual.
Coordination requires thinking about control/communication
interactions. Ensembles of Autonomous Systems
Slide 5
V1V1 VnVn vehicle obstacle control station target start
Trajectory optimization without a map A group of vehicles traveling
between a fixed pair of locations Terrain is unknown - no global
map. On-board sensing provides local information about vehicles
immediate surroundings PROBLEM: Given an initial path between a
pair of start and target locations, find the optimal path
connecting that pair, using local interactions between vehicles. V
n-1
Slide 6
V1V1 VnVn vehicle obstacle control station target start
Trajectory optimization without a map A group of vehicles traveling
between a fixed pair of locations Terrain is unknown - no global
map. On-board sensing provides local information about vehicles
immediate surroundings PROBLEM: Given an initial path between a
pair of start and target locations, find the optimal path
connecting that pair, using local interactions between vehicles. V
n-1
Slide 7
Local pursuit: A biologically-inspired algorithm Theorem (on ):
The iterated paths converge to a straight line as... K+2 k+1 k
Start Target : Initial path : path followed by the k-th vehicle,...
(on a smooth manifold M): If vehicle separation is sufficiently
small, then the iterated paths converge to a geodesic. [Bruckstein,
92]
Slide 8
Experimental results: with Euclidean metric Initial path length
~7m Vehicle separation ~1.5m A collection of mobile robots with:
Wireless communication between neighbors Sonar and odometry sensors
TARGET START
Slide 9
: Minimum-length geodesic connecting to M : location of k-th
vehicle Local Pursuit Idea: Find optimal trajectory to leader and
follow it momentarily.
Slide 10
M : Minimum-length geodesic connecting a to b : location of
k-th vehicle Pursuit decreases vehicle separation
Slide 11
The k-th vehicle moves as follows: Wait at until t=(+1) At time
t, follow the optimal trajectory from to Local pursuit for more
general optimal control problems Let Given an initial trajectory
with converge to a local min. for Assumptions: uniqueness,
smoothness Find that minimizes s.t. As, iterated trajectories
Slide 12
Simulation: pursuit on 5m trajectory 0.7m separation
Slide 13
A sub-Riemannian example 5m trajectory 1.5m separation
fixed
Slide 14
Pursuit in vehicles with drift (minimum time problem)
Slide 15
Summary and Work in Progress A biologically-inspired trajectory
optimization algorithm - local pursuit forms a string of vehicles -
each vehicle uses local information and communicates with its
closest neighbors Target state and optimal trajectory are unknown
Local convergence Experiments Escaping local minima Comparison with
gradient descent methods
Slide 16
Slide 17
Control in a reasonably complex world The problem of specifying
control tasks (e.g. go to the refrigerator and get the milk)
Solving motion control problems of adequate complexity Many
interesting systems evolve in environments that are not smooth,
simply connected, etc. Using language primitives to navigate:
Specify control policies Represent the environment (what parts do
we ignore?)
Slide 18
Def: MDLe is the formal language defined by the context free
grammar with production rules: Motion Description Languages
Evaluate Evolve under until Concatenate, encapsulate atoms to form
complex strings (plans), e.g. N: nonterminals T: terminals S: start
symbol : empty string Fact: MDLe is context free but not regular
Atom:
Slide 19
Symbolic Navigation Keep only interesting details about how to
navigate the world Landmark: L = (M,x)M: map patch, x: coordinates
Sensor signature: L = L i if s(t) = s i (t) for t in [t 0,T]
Navigation Local navigation: on a given landmark L i Global
navigation: between landmarks M x World
Slide 20
A directed graph representation of a map Represent only
interesting parts of the world. G = {L,E} L i : landmarks E ij :
{i,j, ij } ij : an MDLe program E ij E ji Idea: Replace details
locally by a feedback program
Slide 21
Experiment: indoor navigation Lab 1 Lab 2 Office Partial floor
plan of 2 nd floor A.V. Williams
Slide 22
Experiment, contd Goal: Navigate between three landmarks Front
of labRear of labOffice