In order to overcome the motion difficulties of robot near its singularities,a new motion planning algo- rithm is proposed on the basis of task decomposition and speed distribution.In this algorithm,the control task is di- vided into several sub-tasks according to their priorities,and each sub-task is controlled independently.The velocity component of each sub-task is determined according to the driving capability of joint motors and the priority of each sub-task,which helps implement the coordinated movement of robot even in the condition of low flexibility when the robot moves near the singularities or stays in the singularities,thus achieving the control objective and avoiding the vibration,excessive end-effector error and system instability caused by the sudden change of joint relative velocity.By taking a PUMA robot for example,the composition of the control system is illustrated,and the feasibility of the proposed method is proved via simulation.