A Multi-Robot Pattern Formation Algorithm Based On Gradient Descent Technique

Jun Zeng,Xin Zhou,Alei Liang,Haibing Guan

2010-01-01

Abstract:Multi-Robot pattern formation is fundamental and significant in the robotics research community. With the increasing number of robots in one task, team efficiency improvement and energy consumption reduction research becomes more important in robots coordination. In this paper, we propose a fast, efficient algorithm which based on gradient descent technique to the problem of pattern formation on a grid map, for a homogeneous multi-robot system. Basically, we derive map information from the grid map using a gradient descent technique and the map information is proposed as the message passing mechanism among the robots. The robots maintain and modify a map together and make distributed movement decisions through that map information. This method speeds up the formation process and significantly reduces the communication between the robots. Finally, we verify our approach through simulations and real experiments.