動体を扱うロボットの運動計画と制御

概要

論文の詳細を見る
This paper describes robot motion planning in a very dynamic environment, that is, a trajectory generation to attain a working purpose with redundancy freedom to use under the restriction of time and space. In particular, we focus on "juggling" as the typical example of such a task. We will describe a parametric representation of the robot motion which makes it easy to generate the appropriate trajectory of the end-effector for the "juggling" task. Also, we will describe a hierarchical learning control strategy which is effective in generating a desired motion pattern for juggling which has not been previously taught to the robot. For an actual robtot juggling several balls, we predict the location and time at which a ball is caught using sensory feedback signals and modify the end-tip trajectory. Experimental results obtained using a 2-d. o. f. direct-drive manipulator are also presented.
1993-11-25