OPTIMAL AND NEAR-OPTIMAL MANIPULATOR JOINT TRAJECTORIES WITH A PREPLANNED PATH.

Manipulator joint trajectories are sought to optimize a specified cost function under preplanned path constraints, taking into consideration the physical constraints on the kinematics and dynamics of the system. Using a time-scale factor kappa (t) and a set of joint trajectories representing a geometric path, all joint trajectories tracing the path are described as a function of kappa (t) and its time derivative. Then the desirable kappa (t) is obtained by two methods: a global optimization method using dynamic programming (DP), and an iteratively improving (II) method, which is a feasible method and utilizes the local controllability of B-splines. These methods are applied to joint trajectory planning of a preplanned collision-free path of a manipulator. The numerical results show that the II method is more effective than the DP method with respect to computation time and memory requirements if a strictly optimal solution is not required.