Modeling and Precision Motion Control of Dexterous Object In-Hand Rotation

This paper presents the modeling and precision motion control of an important class of dexterous object in-hand rotation by three-fingered robot hand. First, we explicitly construct the planar friction model between the fingertip and the object. Then, to precisely regulate the object angle to its desired value while minimizing the deviation of the object center with respect to the fingertip, we propose a novel hybrid control structure consisting of two modules - regular tracking control module and the motion re-planning module. The proposed approach naturally mimics the human motor behavior when rotating an object to a desired angle by pushing and re-positioning the fingers. Finally, a simulation study is also performed by using a high-fidelity virtual physics engine 'MuJoCo' to verify the effectiveness of the proposed approach.