Aeronautics & Astronautics Autonomous Flight Systems Laboratory All slides and material copyright of University of Washington Autonomous Flight Systems.

Aeronautics & AstronauticsAutonomous Flight Systems Laboratory

All slides and material copyright of University of Washington Autonomous

Flight Systems Laboratory

Aeronautics & AstronauticsAutonomous Flight Systems Laboratory

Research and Development at the

Autonomous Flight Systems Laboratory

University of Washington

Seattle, WA

Guggenheim 109, AERB 214(206) 543-7748

http://www.aa.washington.edu/research/afsl

Aeronautics & Astronautics


University of Washington 3

General Information

Research Focus• Multi-Vehicle Cooperative Control Flight Testing

• Cooperative Strategies for Teams of Autonomous Air & Surface Vehicles

• Probability Based Searching/Target Identification

• Coordinated Underwater Robotics

• Communications for Heterogeneous Cooperating Autonomous Vehicles

To conduct research that advances guidance, navigation, and control technology relevant to Autonomous Vehicles.

Mission Statement

Dr. Rolf RysdykDr. Juris VagnersDr. Uy-Loi LyDr. Kristi MorgansenDr. Anawat Pongpunwattana

Christopher LumCraig HusbyJohn OsborneRichard WiseElizabeth Bykoff

PeopleBen TriplettDan KleinJim Colito




Hierarchy of Autonomy

Path PlanningTask AllocationSearch PatternsHuman Mission Command

Strategic (low bandwidth)

Tactical (medium bandwidth)

State StabilizationSignal TrackingInner Loop or “autopilot”Configuration changes

Dynamics and Control (high bandwidth)

Target ObservationPath FollowingCommunication & CooperationHuman Monitor Interaction




Topography of Autonomous Flight




Hardware-in-the-Loop Simulator

Avionics Tray

HiL Simulator




Hardware-in-the-Loop Simulator

Groundstation Aircraft




Distributed Real Time Simulator

Five computers running REAL TIME simulation software.

Used as a high fidelity testing environment to accurately simulate data transfer and communication aspects.




Infrastructure of Flight Tests

In addition to simulation, direct access to actual hardware and systems.

Partnered with the Insitu Group for ScanEagle UAVs, Northwind Marine for SeaFox Boats.

Extensive test infrastructure in place by working with these local companies

Includes sea launch & retrieval of UAVs




Aspects of Autonomy

Base




Aspects of Autonomy

Base

STRATEGIC Team Assembly Task AssignmentTACTICAL Pattern HoldDYNAMICS & CONTROL Auto Launch/Retrieval




Aspects of Autonomy

Base

Pattern hold/Team assembly




Aspects of Autonomy

Base

TransitionPattern hold/Team assembly




Aspects of Autonomy

Base

TransitionPattern hold/Team assembly

STRATEGIC Path Planning Adaptive Task Assignment

TACTICAL Obstacle/Threat Avoidance Path Following

DYNAMICS & CONTROL State Stabilization




Aspects of Autonomy

Base

Transition

Obstacle/Threat Avoidance





Aspects of Autonomy

Base

Transition



STRATEGIC Dynamic Task Allocation Team-Based Cooperation Path Re- planningTACTICAL Obstacle Avoidance Engagement ManeuversDYNAMICS & CONTROL State stabilization




Aspects of Autonomy

Base

Transition



Coordination w/ surface vehicles




Aspects of Autonomy

Base

Transition

Obstacle avoidance



STRATEGIC Provide improved target tasking

and routing info to unmanned surface vehicles

TACTICAL Orbit Coordination Communication Path FollowingDYNAMICS & CONTROL Signal Tracking




Aspects of Autonomy

Base

Transition




Searching/Target ID




Aspects of Autonomy

Base

Transition

Obstacle avoidance

Coordination w/ ground vehicles


Searching/Target ID

STRATEGIC Map-Based and Probabilistic Searches

TACTICAL Path following

DYNAMICS & CONTROL State stabilization




Aspects of Autonomy

Base

Transition


Searching/Target IDCoordination w/ surface vehicles





Current Research Projects

Real Time Strategic Mission Planning dynamic task and path planning for a team of autonomous

vehicles to cooperatively execute a set of assigned tasks.

Coordination of Heterogeneous Vehicles developing robust navigation and guidance algorithms to

coordinate multiple vehicles to perform a cooperative task.

Autonomous Search and Target Identification using total magnetic intensity measurements to search

and identify magnetic anomalies in a predetermined area.




Real Time Strategic Mission Planning

Base

Transition







System Overview

Previously funded by DARPA & AFOSR




System Block Diagram

Solving optimal control problems in real-time

planstaskD

pathsQ




Stochastic Problem Formulation

Predicted probability of survival of each vehicle at time tq+1

Predicted probability that a task is not completed at time tq+1

Team utility function

ON

j

Oj

Oj

vj

Vv

Vv qqBqq

1

)()1(1)()1(

V TN

v

N

j

vij

Vv

Vv

iv

Fi

Fi dqqBqxqx

1 1

)()1(1)()1(

Mission Score CostJ

Vv

Fix

1

1 1 11

( ) ( ) 1 ( 1) ( ) ( ) ( ) ( ( ))V VT T

p

N NN NNF F i V V v V V V Qi i v v v ij v v p v v v p

i q s j vv

J q x q B q q d s N F Q s




Distributed Architecture for Coordination of Autonomous Vehicles

Each vehicle plans its own path and makes task trading decisions to maximize the team utility function

There is one active coordinator agent at a time efficiency failure detection local/global information

exchanges Computational requirement

for running coordinator agent is small compared to planning

Coordinator role can be transferred to another vehicle via a voting procedure




Evolution-based Cooperative Planning System (ECoPS)

Uses Evolutionary Computation-based techniques in the optimization of trading decision making and path planning

Task planner uses price and shared information in addition to predicted states of the world for making trading decisions

Task planner interacts with path planner and state predictor to simultaneously search feasible near-optimal task and path plans.

We call this system the “Evolution-Based Collaborative Planning System” – ECoPS, combining market based techniques with evolutionary computation (EC).




Evolutionary Computation (EC)

Motivated by evolution process found in nature

Population-based stochastic optimization technique

Metaphor Mapping




Features of Evolution-Based Computation

Provides a feasible solution at any time

Optimality is a bonus

Dynamic replanning

Non-linear performance function

Collision avoidance

Constraints on vehicle capabilities

Handling loss of vehicles

Operating in uncertain dynamic environments

Timing constraints




Market-based Planning for Coordinating Team Tasks

)(,),(),()( 21 nnnn vNTTTA

)(max AAJTask allocation problem:

At trading round n

)()()()1( nSnBnn iiii TT

At the end of the trading round:

The goal of task trading:

))(())1(( nJnJ AA

Each vehicle proposes ( ), ( )i iB n S n

which are approved by the auctioneer

based on bid price.

Distributed Task Planning Algorithm




Dynamic Path Planning

Generate feasible paths and planned actions within a specified time limit (ΔTs ) while the vehicles are in motion.

Highly dynamic environment requires a high bandwidth planning system (i.e. small ΔTs).

Formulate the problem as a Model-based Predictive Control (MPC) problem

1

pp sss ttT




EC-Based Path Planning

MutationDynamic Planning

Path Encoding




Collision Avoidance

Model each site in the environment as a uncertainty circular area with radius

Probability of intersection: use numerical approximation computationally easier than true solution

, ( ), ( )v v Vi i v i i

k

B z k C k v k t

i

vi

: possible intersection region

: probability density field function

: position on the path

Ci : expected site location

v : velocity of the vehicle

viZ

Vvz




Collision Avoidance Example




Simulation Results

Simulation on the Boeing Open Experimental Platform




Some Aspects of ECoPS

Each vehicle computes its own trajectory and makes decision to trade its tasks with other vehicles.

Vehicles may sacrifice themselves if that benefits the team. Each vehicle needs to have periodically updated locations of

nearby vehicles only for collision avoidance. Each vehicle needs to know the information about the

environment. The accuracy of the information affects the quality of its decision making.

The rate of environment information updates should be selected based on how fast objects move in the environment.

Assuming vehicles are equipped with on-board sensors, sharing sensed data improves the performance of the team.




Coordination of Heterogeneous Vehicles

Base

Transition







Coordination and Communication with Autonomous Surface Vehicles

At strategic level, UAVs can provide improved target tasking and routing information to surface vehicles

Autonomous path planning for surface vehicles through non-structured environments enhanced by UAV information

At tactical level, UAVs can track evasive targets and update world estimates

Currently funded under WTC Phase I Fall/Winter ‘05




Goals and Advantages

Goals Use multiple low-cost UAVs to

cooperatively track targets Ability to mark targets, report to

central database, report to deployed surface vehicles

Improve quality and quantity of ISR data and battlefield awareness

Advantages Tracking targets with tactical

UAVs can require high operator workload

Evasive targets could fool a single UAV




Simulation Visualization

Autonomous Orbit Coordination for Multiple UAVs




Simulation Results

Effects of Radius and Airspeed Manipulation




Simulation Results

Effects of Radius and Airspeed Manipulation




Orbit Coordination

Maintains relative phase angle between two UAVs in presence of disturbance

Nonlinear issues dealing with asymmetry of varying orbits

Joint effort between UW, Cornell, U of Calgary, and The Insitu Group

Insitu SeaScan tracking moving target




Autonomous Search and Target Identification

Base

Transition







Probabilistic Searching

Evaluation of Autonomous Airborne Geomagnetic Surveying

Utilize magnetometer to measure local magnetic anomalies for known signature

Identify and classify anomalies

Search for and track anomalies cooperatively

Currently funded under WTC Phase II




General Architecture

Obtaining local magnetic map

Data from Fugro Airborne Surveys




General Architecture

Groundstation

Agent 1

Agent 2

Local Magnetic Map Occupancy Map




Occupancy-Based Map Search

False Anomalies

Target

Agents





Score Cell

Evaluate possible control population

Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm





Score Cell


Execute control

Basic Algorithm




Anomaly Encounter

Aeromagnetic Data from Fugro Airborne Corresponding Line Data

Goal: Classify anomaly as target or false signature

Anomaly

How to score each cell?




Particle Filter

How consistent is trace with trajectory over desired target?

Classify using Particle Filter

Nonparametric Bayes filter. Similar to Unscented Kalman or discrete Bayes filter.

Which trajectory (if any) would produce trace?




Particle Filter

Fox, D., Thrun, S., Burgard, W. 2005, “Probabilistic Robotics”

tx 1,| tttmotion xuxfSample from

ttsensort xzfmw |

for m=1:M

tt xm :,

end

),,(ilterparticle_f function 1 tttt zu

t sampled from t w/probability α tw

Klein, D.J., Klink, J.O., 2005, “Mobile Robot Localization”

tmxtx 2

tx1

t




True Anomaly Encounter




Different Magnetic Signatures

What about for false anomalies?




Confidence Comparison

Actual Target Encounter False Encounter

Features

Use combination of particle filter and neural net to identify target and quantify confidence.




Contact Us

InvestigatorsDr. Rolf Rysdyk [email protected]. Uy-Loi Ly [email protected]. Juris Vagners [email protected]. Kristi Morgansen [email protected]. Anawat Pongpunwattana [email protected]

Autonomous Flight Systems LaboratoryGuggenheim 109(206) 543-7748http://www.aa.washington.edu/research/afsl

Nonlinear Dynamics and Control LaboratoryAERB 120(206) 685-1530http://vger.aa.washington.edu

Aeronautics & Astronautics Autonomous Flight Systems Laboratory All slides and material copyright of University of Washington Autonomous Flight Systems.

Documents