In order to address the problems of low efficiency and high randomness of the RRT-based arm path planning algorithm based on random sampling in the global workspace, a mechanical arm path planning algorithm based on sample point optimization RRT algorithm is proposed. Compared to global workspace sampling, the optimization algorithm first generates random sampling points based on non-obstacle space to reduce the probability of algorithm collision detection and redundant node generation, and combines with a certain probability of artificial potential field method to generate heuristic sampling points, which can ensure the completeness of random sampling probability and make the sampling points more target-oriented. Secondly, in order to make the path more concise and smooth, a redundant node deletion strategy is used to remove redundant nodes in the path to optimize the final path. Finally, comparative experiments and analysis of the optimized algorithm are carried out in two-dimensional and three-dimensional simulation environments, confirming the good performance of the algorithm in random sampling path planning algorithm, and conducting obstacle avoidance planning algorithm experiments on the IRB 1200-7/0.7 robot arm. The simulation and experimental results both demonstrate that this algorithm can achieve higher planning efficiency and better path in robot arm path planning.