The paper proposes the dynamic path planning method of mobile robot based on improved ge- netic algorithm. Firstly, to ensure the safety of the robot e®ectively, the improved visual graph and the conception of safety coe±cient are presented. Secondly, to avoid the local optimum, the paper replaces the mutation individual with a better individual searched by hill-climbing method. To increase the convergence speed, the paper updates the colony by the rule of the particle swarm optimization. Thirdly, °oat-point coding is proposed in the improved ge- netic algorithm. The ¯tness function includes the length and the total slope of the path in dynamic path planning. Finally, it emulates the dynamic path planning for mobile robot based on improved genetic algorithm. From the re- sults, it can see that the path planning method is viable and e±cient.