An Intelligent Force Control Strategy for Soft Robotic Grippers
University of New Brunswick
Researchers are striving to develop more human-like robots, and a stepping stone towards this goal is to mimic the human grasp . Many traditional robotic grippers are application specific with rigid fingers and singular force and position trajectories. This work explores a proof of concept for an electric two-finger parallel gripper that is able to accommodate a variety of objects with different sizes and weights and accurately control the gripping forces in a manner similar to humans. The scope of this thesis was to develop a methodology to identify objects, predict the size and required gripping force and subsequently control the gripping force for a subset of object classes. The thesis exploits object recognition algorithms, computer vision techniques and traditional control schemes in order to meet these objectives. The methodologies were demonstrated in simulation and experimentally. Intelligent and accurate gripping control was achieved.