Interactive Shepherding

Jyh-Ming Lien and Emlyn Pratt


Shepherding behaviors are one class of flocking behaviors in which one or more external agents (called shepherds) attempt to control the motion of another group of agents (called flock) by exerting repulsive forces from shepherds to the flock.

In this work, we studied how to integrate interactive methods with motion planning to facilitate better shepherd navigation.
We used laser pointers and a vision-based tracking system to control the motion of the shepherd.
Our goals include: studying (1) how user input can help shepherd navigates the flock, (2) how the motion planner can provide
useful hint for guiding users, (3) how the motion planner can learn from the interactions and improve itself.

Interactive Planning for Shepherd Motion, Jyh-Ming Lien and Emlyn Pratt, Proceedings of 2009 AAAI Spring Symposium, AAAI, 2009
Paper (PDF) / BibTeX

A snapshop of our system is shown in the figure below.


This video shows the system in action

See Also

List of MASC Research Pages
Computer Science @ George Mason University