Force Control in Flexible Object Grasping Based on Position Predictor

A novel manipulation method for grasping flexible objects with a simple rigid robot hand is proposed in this study. Based on a position predictor, it mainly focuses on force control during the object grasping. A desired force for grasping is given at first, then during the process of grasping it is converted to a predicted displacement. An online linear approximator for solving the predicted displacement iteratively is proposed. Here Model Predictive Control (MPC) is adopted to achieve accurate position control. Meanwhile a flexible model is designed for the simulation and verification. This method is verified in a flexible object contact situation and the result indicates that it can control the force precisely in flexible object grasping and achieve better performance compared with admittance control.