Multi-robot formation reconfiguration is an important subject in multi-robot systems, especially for applications of formation control of multiple UAVs and multiple UUVs. Under the support of the National Basic Research Program, multiple robot formation reconfiguration based on path planning is investigated. Firstly, in this thesis, we analyzed and concluded the constraints and optimal objectives, such as mechanism constraints, environment constraints, multi-robot coordination control constraints and shortest length, minimal energy consumption, which should be considered in multi-robot formation reconfiguration and path planning. Secondly, a path planning method based on Bezier curve for a single robot is presented. And a feasible path from original position to target position within constraints of velocity, acceleration, turning radius can be generated to make the robot arrive at the target position at a given time. In addition, the proposed method is also suitable for robots to plan paths to avoid obstacles. Thirdly, a method for multi-robot formation reconfiguration based on the path planning method is proposed. The proposed algorithm can assign a target posiont for each robot in a new formation automatically, and generate paths within constraints of multi-robot collision avoidance, obstacle avoidance and simultaneous arrival for the robots to form the new formation definitely. Finally, a simulation platform is developed with GUI and M-file program tools of MATLAB. The simulation platform can easily fulfill path planning for multi-robot formation reconfiguration with the algorithms in this paper.