Fast Formation Transformation and Obstacle Avoidance Control for Multi-agent System

In this paper, formation transformation and obstacle avoidance problems of agents passing through complex environment are discussed. Firstly, the formation dynamic transformation is realized based on the time window prediction mechanism according to the changes of the surroundings. Secondly, to speed up the formation transformation and recovery, this paper introduces the Hungarian algorithm into the dynamic scene, and comprehensively considers the distance, angle, crossing cost between the agent and the target point to achieve the optimal allocation. Finally, the dynamic window approach (DWA) is improved by adding target guidance and combined with consensus algorithm to achieve safety of obstacle avoidance. The simulation results show that the proposed formation allocation algorithm can effectively reduce the formation transformation cost and recovery time and can restrain the conflict among agents. In addition, the proposed obstacle avoidance algorithm improves the smoothness of agent's trajectory, and reduces the damage to the formation shape during the obstacle avoidance process, which has contribution to the rapid recovery of the formation.