Slip Prediction Based on Manipulator Motion

Abstract

Predicting slip during robotic manipulation is of interest for a variety of appli-cations. Especially applications where weak grasps are applied. In this thesis, amodel for predicting slip for a two fingered grasping scenario is considered. Otherthan model parameters, the only measurements or sensor information assumed isof the manipulator joints. Soft objects that deform substantially under appliedforces are especially interesting in terms of frictional behaviour. A soft ball wasused as a test object and parameters for friction and deformation was experimen-tally determined. By grasping and moving the ball with an industrial manipulator,slip and object loss was induced in order to compare these observations againstmodel predictions.It was found that the models prediction of slip was reasonable when compared tothe observations. However, the model could not be fully tested and validated be-cause the simple geometry of the test object did not excite any frictional behaviourfrom the soft characteristics.