AN ONLINE ROBOT PLANNING STRATEGY FOR TARGET INTERCEPTION

On-line planning of gross robot motion for moving-object interception is considered in this article within the context of an active predicti on planning, and execution (APPE) strategy. The objective is to find a n optimal interception point such that the robot end-effector and the object arrive simultaneously at this target pregrasping point. In this approach, the optimality of the selected rendezvous point on the targ et trajectory is directly dependent on the robot-trajectory planning t echnique. Thus, for the solution of the general interception problem, three issues must be addressed: (i) optimal rendezvous-point selection , (ii) optimal robot-trajectory planning, and (iii) replanning in resp onse to gross changes in the predicted target trajectory. The effect o f uncertainties in the target-trajectory prediction must be considered at each planning stage. Herein solutions to the first two problems ar e briefly reviewed as background to the proposed rendezvous-point repl anning strategy. This strategy determines when replanning is necessary modifies the rendezvous point, and iteratively replans robot ''patch trajectories'' to new interception points. Simulation results using tw o different on-line robot-motion-generation strategies are also presen ted. (C) 1998 John Wiley & Sons, Inc.