This paper presents an improved A*algorithm and it is applied to the robot path planning. Adopting the twice-path-planning strategy,the robot can re-plan its path effectively when falling across an unknown obstacle suddenly in the given environment. Adopting the child node generation strategy based on priority, the path planning can be implemented in reality dealing with the volume of the robot. The experimental results show the validity and reliability of the algorithm in MATLAB simulator. The improved A*algorithm improves the robot’s intelligence and real-time path planning ability.