Dexterous robotic hand grasp modeling using piecewise linear dynamic model

摘要

Learning from sensor data is important in many robotic research areas, such as dexterous robotic hand grasping. In this paper, a piecewise linear dynamic model is proposed for analyzing robotic hand grasp. The combination of linear dynamic model and the switched systems can achieve better results in grasp learning due to its advantage of modeling multi-phase grasping process. To the best knowledge of the authors, this is the first time for piecewise linear dynamic model to be incorporated into the framework of modeling robotic hand grasp process. The performance of the proposed model is evaluated on our experimental system and shows promising results.