A New Scheme for Cooperative Hunting Tasks with Multiple Targets in Dynamic Environments

Abstract

Multiple robot cooperation has been playing more and more significant role in scientific and engineering applications. The cooperative hunting task with multiple targets is a critical issue in multiple robot problem. To successfully hunt multiple preys, predator robots not only need to settle the problem on how to cooperatively hunt a prey, but also find an efficient task allocation to assign each predator to hunt certain prey. This paper proposes a cooperative hunting scheme based on the multi-target k-winner-take-all (k-WTA) algorithm and the wolf-pack-particle-based model. Based on the competitive mechanism, the multi-target k-WTA algorithm is used for task allocation with an optimized hunting strategy and achieves the allocation within a short period of time in dynamic environments. With simple rules, the wolf-pack-particle-based model can trigger collective behaviors and enable predator robots to encircle the prey. The simulative experiment results show that the proposed approach is capable of efficient task allocation and guiding predator robots to round up multiple moving targets in dynamic environments.