Multi-Agent Coverage Path Planning via Proximity Interaction and Cooperation
Lei Jiao,Zhihong Peng,Lele Xi,Shuxin Ding,Jinqiang Cui
DOI: https://doi.org/10.1109/jsen.2022.3150098
IF: 4.3
2022-03-15
IEEE Sensors Journal
Abstract:In multi-agent systems, the decision of an agent will be affected by the behaviors of others. Therefore, from the perspective of an agent, the situation is uncertain and random. Inspired by the social behaviors in the biological world, a novel multi-agent coverage path planning algorithm is proposed. Based on the positions of agents, the problem is decoupled, which can effectively reduce the dimension of the decision space. The behavior-guide-point is introduced to guide agents in making decisions, and a new motion mode is presented. To avoid falling into the local optimum, a cooperation mechanism is designed, which can improve the adaptability of the system. Through proximity interaction, the prediction results obtained via the model predictive control (MPC) technology are fused, evaluated, and sorted within the neighborhood, based on which decisions are gained. The proposed algorithm can handle emergencies in unknown environments such as body damage and moving obstacles, and can also be applied to heterogeneous systems. Simulation shows that compared with other algorithms, it has advantages in terms of the makespan and the coverage repetition rate.
engineering, electrical & electronic,instruments & instrumentation,physics, applied