An active prediction, planning and execution approach to object interception is applied to the problem of catching a moving object. The object motion cannot be accurately predicted into the future necessitating active path replanning. Unlike the APPE systems reported in the literature, which at best can select a rendezvous-point amongst a few non-optimal interception points, the proposed APPE system is able to intercept the object anywhere along its predicted trajectory. Time-optimal object interception is accomplished by planning the rendezvous-point at the earliest possible time using a quintic-polynomial trajectory planner. Simulation and experimental results are reported.
展开▼