Predicting a moving object position for visual servoing: Theory and experiments (Q2731545)

When a robot manipulator works in an unknown environment, it is necessary to use sensors such as video cameras or force and tactile sensors. The problem which then occurs is the length of the time lag introduced by the corresponding image processing, and one way to cope with this trouble is to estimate the trajectory of the moving object to be manipulated. NEWLINENEWLINENEWLINEIn the present note, the authors compare the performances of three prediction algorithms: the first one is based on Kalman filtering, and the two other ones are adaptive predictors which use gradient and least-squares estimation schemes. Kalman filter predictor performances are affected by speed changes on the moving object but are less expensive in terms of computational burden.