A Novel Path Planning Algorithm for Multi-Agent Collaboration in Finite Space

Multi-agent collaboration can complete tasks that are difficult for a single agent to handle, making a big difference on the battlefield, in healthcare, and in other domains. Effective path planning is an important factor in ensuring task execution. However, the finite space increases the probability of multi-agent collisions, which poses a challenge to multi-agent path planning. To this end, we propose a novel path planning algorithm by two-module fusion (PLA-TMF). Specifically, the first module is to develop a collision detection model based on discrete point intersection probabilities, which improves the collision detection accuracy between different agents. Then, the second module is to realize the path planning algorithm under the minimum collision probability constraint, which relies on genetic algorithm. Finally, the effectiveness of the proposed algorithm is verified by simulation experiments under specific scenarios.