Aiming at the problems of low search efficiency of FMT* algorithm in robot arm path planning and easy planning failure when there are fewer sampling points, this paper proposes a US-FMT* algorithm based on uniform sampling to improve the planning success rate and search efficiency. The algorithm first adopts a uniform division of the global workspace to generate random sampling points, which improves the performance of the algorithm under low sampling point conditions; then, it is combined with the neighbor node search-oriented strategy to reduce redundant node exploration in the planning process; finally, it adopts the node reconnection strategy to optimize the quality of the path. The algorithms are experimentally analysed in 2D and 3D simulation environments, and path planning experiments are performed on an ABB IRB 1200-0.7/7 kg robotic arm. In the 3D path planning simulation environment, the US-FMT* algorithm reduces the time cost by about 49.7% and the path cost by about 16.6% compared to the FMT* algorithm, which shows that the US-FMT* algorithm is able to effectively improve the planning success rate of robotic arm path planning, provide excellent path quality and lower computational cost, and provide an efficient solution for robotic arm motion planning.