Abstract: This paper proposed an improved RRT algorithm for the path planning problem of the seven degrees of freedom（7-DOF）mechanical arm. Based on the RRT algorithm，optimization strategies such as two-way search，greedy idea，dynamic step size，and new node generation were added to shorten the algorithm's running time and path length. Under ROS，the RRT algorithm was simulated and compared with the improved RRT algorithm. The results showed that compared with the RRT algorithm，the planning time of the improved RRT algorithm was shortened by 65. 1%，the path length was shortened by 50. 6%，and the success rate was increased by 4. 0%，which achieved the expected effect.

Key words: rapidly-exploring random tree（RRT）, ROS, mechanical arm, path planning