Real-time trajectory generation using model predictive control

Abstract

The problem of planning a trajectory for robots starting in an initial state and reaching a final state in a desired interval of time is tackled. We consider Model Predictive Control as an approach to the problem of point-to-point trajectory generation. We use the developed strategy to generate trajectories for transferring the state of the robot, fulfilling computational real-time requirements. Experiments on an industrial robot in a ball-catching scenario show the effectiveness of the approach.