抄録
This paper discusses how to plan manipulator joint trajectories in order to make a performance index as small as possible under geometric path constraints in consideration of physical contraints based on kinematics and dynamics of a mainpulator system. Using a time scale factor κ(t), the specified index and joint trajectories are described by functions of κ(t) and its time derivative κ(t). Then the desirable κ(t) is obtained by two proposed methods: the first one is the iteratively improving (II) method which is a feasible method and utilizes the local controllability of B spline; and the second one is a global optimization method using dynamic programming (DP). These methods are applied to an example of a collision-free path of the manipulator with four links and four degrees of freedom. The numerical results point out that II method is much more effective than DP method in view point of the computation time and used memories, if the strictly optimal solution is not required.
