Title: Cooperative Pathfinding based on high-scalability Multi-agent RRT*

Abstract: Problems that claim several agents to find no-conflicts paths from their start locations to their destinations are named as cooperative pathfinding problems. This problem can be very challenging and it has received a lot of attention. There are many works proposed to solve this problem in the last decades: coupled, decoupled, and hybrid approach. In the field of coupled approach, the computational cost of many algorithms increases as the dimension of problems increases. Recently, an algorithm called Multi-agent RRT*(MA-RRT*) was proposed. This algorithm can alleviate the increase in computational cost and efficiently solve cooperative pathfinding problems. However, in a relatively dense environment, the application of MA-RRT* is hindered, because some of the random samples in the free space cannot be explored by the rapidly random tree. This paper proposes an improved version of MA-RRT*, called Multi-agent RRT* Potential Field (MA-RRT*PF), an anytime algorithm that can efficiently guide the rapidly random tree to the free space in relatively dense environments. It works by incorporating a potential field to the GREEDY function to enhance the ability to avoid the obstacles. The results show that MA-RRT*PF performs much better than MA-RRT* in relatively dense environments in terms of scalability while still maintaining the solution quality.