Shepherding2ndProgressReport

Transcript

1. 2nd Progress Report Selection of My Shepherding Research Method Aiyi LI Wakamiya Lab li-aiyi@ist.osaka-u.ac.jp 1 August 6th, 2021
2. Explain more with my research background and the reason for picking up that specific method “Related works” section in paper draft Motivation 2
3. Flocking, steering behaviors Shepherding in single shepherd agent Coordination between multiple shepherd agents Sequences 3 Criticism included
4. Flocking arises from simple rules and does not involve any central coordination. The Boid model is the first model to describe the motion of flocking behaviors. Flocking behaviors 4 Flow field following steering behaviors F: Flow direction S: Steering direction V: Current velocity Reynolds, Craig W. "Steering behaviors for autonomous characters." Game developers conference. Vol. 1999. 1999.
5. Single shepherd agent (Prior methods) 5
6. Steering the whole flock Driving the target sheep agent (Instead of flock) Single shepherd agent 6
7. Selecting the nearest neighbour Driving the target agent 7 Lee, Wonki, and DaeEun Kim. "Autonomous shepherding behaviors of multiple target steering robots." Sensors 17.12 (2017): 2729. Thoughts: We can inspire nearer-neighbour method from this nearest neighbour strategy Shepherd velocity partly depend on the position of the nearest sheep agent
8. Defects for single agent Success is not ensured when conditions complicated Difficulties for multiple agents Coordination between shepherds From single to multiple agents 8
9. 9 Multiple shepherd agents (Prior methods)
10. Obeyed to outer instructions Moving together in a queue-like shape Behaving separately but with messages transferred Multiple shepherd agents 10
11. Behaving separately but with messages transferred Multiple shepherd agents 11 Lee, Wonki, and DaeEun Kim. "Autonomous shepherding behaviors of multiple target steering robots." Sensors 17.12 (2017): 2729. Current flock: Main flock Scattered flock: Other smaller flock Messages transferred The positions between other shepherd agents and sheep agents Thoughts We can inspire from the type of messages to obtain more distribution.
12. Our methods 12
13. Single: Nearer neighbors Multiple: Behaving separately with messages transferred Evaluation: Simple but strong Points: Distributed cooperation, locale factors and robustness Our methods 13
14. Timeline 14 Time Pace early August Paper writing mid August - September Internship, Paper writing October Autumn semester