An improved real-time tracking scheme based on gradient projection for kinematically redundant manipulators

摘要

Gradient projection is the common inverse-solution method for kinematically redundant manipulators during realtime tracking the planned path in obstacles environment. However, the end-effector of kinematically redundant manipulators usually cannot achieve an ideal tracking of the planned path due to the calculation errors. This paper presents an improved real-time scheme based on the gradient projection for redundant manipulators to realize the accurate kinematically tracking in multi-obstacles environment. The real-time path is planned according to the shortest path principle. Gradient projection method is used to calculate the inverse kinematics on the basis of the planned path in real time. The strategy of fixing joint angles is proposed to restrict several joint motion by considering the priorities of the distance between manipulator and obstacles, and the singular position. The unrestricted joint angles are compensated by secondary calculation method to reduce the errors caused by gradient projection method. Then, the accurate tracking can be achieved and the object is thus grasped in multi-obstacles environment. The simulation results based on an 8-DOF planar manipulator show that the proposed algorithm is feasible and owns accurate tracking ability.