Cooperative flocking and learning in multi-robot systems for predator avoidance

摘要

In this paper we propose a hybrid system that integrates reinforcement learning and flocking control in order to create an adaptive and intelligent multi-robot system. First, we present a flocking control algorithm that allows multiple mobile robots to move together while avoiding obstacles. Second, we propose a distributed cooperative learning algorithm that can quickly enable the mobile robot network to avoid predator/enemy while maintaining the network connectivity and topology. The convergence of the cooperative learning algorithm is discussed. As a result, the hybrid system of flocking control and cooperative reinforcement learning results in an efficient integration of high level behaviors (discrete states and actions) and low level behaviors (continuous states and actions) for multi-robot cooperation. The simulations are performed to demonstrate the effectiveness of the proposed system.