Abstract: The RRT^* algorithm is an important tool in robotic arm path planning. However, its application in high-dimensional spaces suffers from low search efficiency, high sensitivity to dimensions, and difficulty in rapidly converging to an optimized path. Additionally, the planning for obstacle avoidance by robotic arms requires considering the smoothness of paths, but the paths generated by the algorithm often lack the required smoothness, making them challenging for direct application in practical robotic arm operations. Addressing these challenges, the study proposes an improved version of the RRT^* algorithm based on a greedy strategy. The new algorithm improves the cost function and reconnection strategy and, in high-dimensional search environments, employs a biased sampling approach through a greedy algorithm to adaptively select predefined path nodes, thereby enhancing search efficiency, trajectory smoothness, and direct application. Through Matlab, ROS simulations, and practical obstacle avoidance experiments with robotic arms, the study validates the efficiency and superiority of the improved RRT^* algorithm in three-dimensional space, especially in terms of search efficiency and path smoothness.