Abstract: When the coal gangue sorting robot works, the gangue in the coal gangue flow is mainly sorted out through the mechanical arm equipped with an end effector, and the coal gangue flow moves with the belt conveyor, and the objects around the coal gangue to be sorted have a certain blocking effect on the sorting process. Most of the existing manipulator trajectory planning algorithms aim at the end effector to quickly approach the target gangue or safely reach the target gangue, but fail to take into account the safety and sorting efficiency of the manipulator in the dynamic environment. When using a robotic arm to sort a moving mass object such as coal gangue, synchronous gripping should also be considered to reduce the shock load caused by the speed difference between the end effector and the target. In order to ensure the safety of the gangue sorting robot, improve the sorting efficiency and avoid the impact load damage to the end effector, it is necessary to carry out reasonable motion planning for the robotic arm. A coal gangue sorting robot with a three-degree-of-freedom Cartesian coordinate manipulator is taken as the research object, and a trajectory planning method for the obstacle avoidance trajectory of the coal gangue sorting robot based on improved and improved RRT* is proposed. Firstly, according to the coal gangue sequence obtained by the visual recognition system, the sorting income function was used as the index to assign the gangue task to the designated robotic arm, and the obstacle was appropriately expanded according to the size of the end effector, and the remaining coal and gangue in the time window were established as the obstacle environment, and the synchronous tracking constraint relationship between the end effector and the dynamic gangue was established. Then, the artificial potential fields method was used to guide the growth direction of the new nodes in the path expansion process of the RRT* algorithm, and the directionality of the expansion was improved, and an Environment-sensitive target bias strategy was proposed to solve the problem that the artificial potential field method is easy to fall into local optimum. According to the potential field of the new node, the target bias probability threshold is adjusted in real time, the local characteristics of the path are improved, and the improved RRT* (AE-RRT*) is used to solve the obstacle avoidance path from the starting point to the target gangue in the obstacle environment. Finally, considering the synchronization constraints and the velocity and acceleration constraints of each joint of the manipulator, the third-order B-spline curve was used to track the dynamic gangue trajectory planning. Experimental results show that compared with the existing obstacle avoidance trajectory planning methods of coal gangue sorting robots, the path length of the proposed AE-RRT* is reduced by 18.97% and 3.09%, respectively, and the path length and path planning time are reduced by 5.92% and 67.66% compared with the traditional RRT* algorithm. Compared with the existing methods, the proposed obstacle avoidance trajectory planning method for the coal gangue sorting robot can improve the position and speed accuracy of the end of the robotic arm by 33.87% and 51.88% on average while ensuring the sorting efficiency, and the moving obstacles are successfully avoided in the process of sorting dynamic coal gangue and its own operation is stable.