Vision-Based Cyclic Pursuit for Cooperative Target Tracking

A cooperative target tracking that is inspired by the idea of cyclic pursuit was proposed. Cooperative target-tracking control laws are presented for both single-integrator and double-integrator robot models when the target's motion information is known. If the target's motion information is unknown, vision-based estimation schemes are applied, assuming that each robot has monocular vision. If the target's motion is unknown (piecewise) constant, estimation of the target's velocity is performed locally on each robot, independently of other robots. For unknown time-varying target motion, the leading agent also sends its visual measurements of the target to its follower, allowing its follower to form a stereo-vision-type estimation. The effectiveness of the proposed cooperative target-tracking control laws and their vision-based counterparts is demonstrated by numerical simulation examples using nonholonomic robots.