In space, the movement of the space robot is subjected to the law of angular momentum conservation as a nonholonomic constraint. Hence, different trajectories of a manipulator result in different final states in inertial space. To ensure safe and reliable operations of a space robot, it is necessary to plan the trajectory of the manipulator mounted on a space robot before the operations. In this paper, the trajectory planning method using a neural network for the manipulator of a space robot is proposed. The trajectory obtained by this method is compared with that obtained by analytical method for the approximate model in which the mass of the manipulator is neglected. The method is applied to the exact model in which the mass of the manipulator can not be neglected.
展开▼