Abstract:

According to current shepherding behavior, motions of the shepherds lack of the restrictions of time and velocity. An improved method is presented to solve this problem. By combining multi-agent decoupled planning based on probabilistic roadmap and timevarying planning based on velocity-tuning method, the motions of the shepherds satisfy the restrictions. Communication capacity among shepherds is also realized by means of priorities. Thus, a more realistic shepherding behavior is achieved. Simulation tests show that the improved shepherding behavior accords with the characteristics of the real people in riot events more ideally.