behavior-based formation control for multi-robot teams tucker balch, and ronald c. arkin
Post on 19-Dec-2015
221 views
TRANSCRIPT
![Page 1: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/1.jpg)
Behavior-Based Formation Control for Multi-robot Teams
Tucker Balch, and Ronald C. Arkin
![Page 2: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/2.jpg)
Introduction Main idea: Behavior-based approach to robot
formation-keeping. The robots in this research have
different IDs: heterogeneous robot. The formation behaviors were implemented as
motor schema within AuRA, and as steering and speed behaviors within UGV demo II architecture.
![Page 3: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/3.jpg)
Background Why formation:
Formation in nature:1. To detect predators2. To efficiently forage for food Grouping Multiple sensor
Formation in robot: • Robotic scouts also benefit by directing
their sensors in different areas
![Page 4: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/4.jpg)
Craig Reynolds pioneering work: • A simple egocentric behavioral model
for flocking• The behavior consists of several
separate components: 1) collision avoidance 2) Velocity matching3) Flock centering
![Page 5: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/5.jpg)
Tu and Terzopoulos:• Realistic simulated fish schooling.
Brogan and Hodgins:• System for realistically animating herds
of one-legged agents using dynamical models of robot motion.
![Page 6: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/6.jpg)
Wang:• Strategy for robot formation.• Individual robots are given specific
positions to maintain relative to a leader or neighbor.
• No obstacle avoidance and no navigation. Chen and Luh:
• The analysis of group dynamics and stability.
• No obstacle avoidance and no navigation.
![Page 7: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/7.jpg)
Mataric:• Investigated Emergent group behavior.• Simple behaviors can create an emergent
flocking behavior in group• Related to Reynols’ work
Parker:• Coordination of multiple heterogeneous
robots.• “Bounding overwatch”
![Page 8: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/8.jpg)
II. Approach a) Line
b) Column
c) Diamond
d) Wedge
![Page 9: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/9.jpg)
Formation maintenance in two steps:• Detect-formation-position: Determines
robot proper position based on current data.
• Maintain-formation: Generates motor command to correct robot location.
AuRA case: Motor schema control, The command is movement vector.
UGV Demo II: Separate votes are cast for steering and speed correction
![Page 10: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/10.jpg)
1) Unit-center- referenced
2) Leader-referenced
3) Neighbor-referenced
Formation position determined by various referencing techniques
![Page 11: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/11.jpg)
Each robot determines the positions of its peer by direct perception of other robots. GPS, Dead-reckoning
The robot transmit their current position in world coordinates with updates as rapidly as required for the given formation speed and environmental condition.
Position errors and latency are significant factor.
![Page 12: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/12.jpg)
Nomad 150 simulation• Motor schema implementation.
UGV implementation• Decoupling of motor control into separate
steering and speed behaviors.
![Page 13: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/13.jpg)
III Motor schema-based formation control Motor schemas:
• Move to goal, avoid obstacle, avoid robot, noise, and maintain formation.
Determined robot behavior:• Multiplying the outputs of each primitive
behavior by its gain, then summing and normalizing result.
![Page 14: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/14.jpg)
![Page 15: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/15.jpg)
Maintain-formation schema generates a movement vector toward the desired position.
The vector magnitude depends on how far the robot is away.
Ballistic zone, controlled zone, and dead zone
0%10%20%30%40%50%60%70%80%90%
100%
![Page 16: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/16.jpg)
The deadzone provides a stable target area. Mobile robots require a small deadzone to
avoid oscillations about the formation position due to latency in communication.
The formation behavior is one component of the robots’ overt actions.
The resultant action depends on the relative strength of the formation behavior to the other goal-oriented behaviors.
![Page 17: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/17.jpg)
A. Motor schema results in simulation Georgia Tech’s mission lab robot
simulation. Five sided polygons: robot Black circles: obstacle Solid line: Robot path Maintain position process:
Obstacle information request of that data send to simulation process A list comprised of angle and range data returned A robot moves and maintains the formation
![Page 18: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/18.jpg)
Four robots in leader-referenced
![Page 19: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/19.jpg)
Comparison of leader-referenced and unit-center-referenced diamond formation
![Page 20: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/20.jpg)
B. Motor schema performance in turns Evaluation in turns:
• The robots commanded to travel 250m, turn right, then travel another 250m
• 90 degree turn• No obstacles• 10 simulations• Robot travel 100m before the evaluation
starts• 500m course evaluation excluding the 100m
initial travel
![Page 21: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/21.jpg)
Path length ratio: (average distance traveled by the four robots)/(the straight-line distance of the course)
Position error: Average displacement from the correct formation position thru the run
Time out of formation: The percentage time out of formation data
![Page 22: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/22.jpg)
C. Motor schema performance in an obstacle field Performance also measured for four
robots navigating across a field. Obstacles are placed randomly so that
2% of the total area is covered with 10-15m diameter obstacles.
Data from runs on 10 random scenarios were averaged for each data point.
![Page 23: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/23.jpg)
The best performance: column formation In most instances, unit-center-referenced formations fare
better than leader-referenced formation Overall path length for robots in a leader-referenced
formation is generally longer than in unit-centered formation
![Page 24: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/24.jpg)
D. Motor schema results on mobile robot
Two-robot team of Nomad 150 robots used: Shannon and Sally.
The robot were directed to navigate from West to East across the room.
Separate runs were conducted for each type of formation with and without obstacle.
![Page 25: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/25.jpg)
Traveling in column formation
![Page 26: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/26.jpg)
Traveling in wedge formation
![Page 27: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/27.jpg)
Traveling line formation
![Page 28: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/28.jpg)
IV. Formation control for the UGV demo II architecture UGV Demo II is aimed at fielding a
robotic scout platoon for the Army. Each UGV is a High Mobility
Multipurpose Wheeled Vehicle (HMMWV).
The UGV differs from the motor schema method.
• The behaviors are coordinated by speed and turn arbiters.
• Separate motor behaviors are developed for the speed and turning component of the behavior.
![Page 29: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/29.jpg)
Non-holonomic constraint robot’s heading during formation impacts its ability to remain in position.
Positioning:• Fore-aft correction:
adjust speed• Side-side correction:
adjusting heading
![Page 30: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/30.jpg)
Speed selection:1) If the robot is in formation, the best speed for
maintaining that formation is the current speed.2) If the vehicle is behind its position, it should speed up3) If the vehicle is in front of its position, it should slow
down.4) The selected change in speed should depend on how
far out of position the robot is.5) Since the speed arbiter implemented in the Demo II
architecture selects the lowest speed vote of all the active behaviors for output to the vehicle, formation speed control is only possible by slowing down.
![Page 31: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/31.jpg)
For steering:1) If the robot is in formation, the best heading for position
maintenance is the formation axis.2) If the robot is out of position laterally and the formation
is moving, it should turn toward the formation axis with an angle that depends on how far out of position it is.
3) If the robot is out of position and the formation has stopped moving, the robot should head directly toward its postion.
![Page 32: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/32.jpg)
Rpos,Rdir=robot’s present position and heading; Rmag=robot’s present speed; Fpos=robot’s proper position in formation; Faxis=formation’s axis Hdesired=desired heading dheading=computed heading correction dspeed=computed speed correction Vsteer=steer vote, representing the directional
output of the motor behavior. Vrmsspeed=speed vote, the speed output of the
motor behavior.
![Page 33: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/33.jpg)
A. UGV behaviors for formation Demo II architecture requires a decomposition
of control into separate steering and speed control components.
The maintain-formation-speed:Vspeed = Rmag + K*dspeed (Rmag = robot present speed)
• Ballistic zone: dspeed=1.0• Controlled zone: dspeed=1.0~0.0• Dead zone: dspeed=0.0
![Page 34: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/34.jpg)
The maintain-formation-steer:Hdesired = Fdir – dheading
Hdesired = Fpos – Rpos
Vsteer = Hdesired – Rdir
The magnitude of dheading determined as:
• Ballistic zone: 90 degree• Controlled zone: 0~90• Dead zone: 0
![Page 35: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/35.jpg)
speed correction heading correction
![Page 36: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/36.jpg)
V. Results for UGV demo II mobile robots
The behavior were implemented using Single-robot simulator.
The behavior were debugged by generating Artificial fixed trajectory for one vehicle, then observing simulated robot’s attempt to maintain position.
![Page 37: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/37.jpg)
VI. Conclusion Reactive behaviors for four formation and
three formation reference type were presented
AuRA implementation is conceptually simpler and applicable to holonomic robots.
UGV implementation addresses the additional complexity of nonholonomic vehicle control.
In most cases, unit-center ref formations perform better than leader-ref.
![Page 38: Behavior-Based Formation Control for Multi-robot Teams Tucker Balch, and Ronald C. Arkin](https://reader035.vdocuments.mx/reader035/viewer/2022062516/56649d3a5503460f94a14c09/html5/thumbnails/38.jpg)
Thank you!!!Have a nice weekend!!!