Robot Motion Planning for Hand-Over with Human.

摘要

In this paper, we discuss about motion planning for hand-over between human and robot. In constructing an algorithm to generate receiver's motion, we aim at reproducing a human-like robot motion in order to reduce human physical and mental burden. First we analyze trajectories and velocity patterns of a hand-over motion performed by two humans. From this experimental results we extract some conditions to realize safety and kind receiver's motion. We then plan the robot's motion considering these conditions based on an instantaneous optimal control method. This algorithm updates the performance function evaluating the error and torque energy at each sampling step. Therefore we can design the weight matrix concerning the error as time-varying form to bring the characteristics of robot motion close to that of human motion. Finally, we execute some kinds of simulation which generates robot receiver's motion by inputting real human deliverer's data into proposed algorithm. These results shows that the robot motion realize safety and kind receiver's motion reflecting human receiver's characteristics.