An improved dynamic Grid-based potential field method was proposed based on the consideration that the goal, robot and obstacles in robot soccer compete are all dynamic. We combined the advantages of potential field method and the grid method, set the grid method to represent the environment, and got dynamic potential function in the potential field method. We used dynamic potential function to form the inspire function of the search algorithm A* which is used for the search of adjacent nodes. The dynamic Grid-based potential field method meets the real-time planning requirements in the complex and dynamic environment. And it has received very good results in solving the local minima problem of the traditional potential field and improving the planning efficiency. It is better in security and reliability. Simulation results show that the method is feasible and effective in soccer robot path planning.