Two recurrent neural networks for grasping force optimization of multi-fingered robotic hands

摘要

Two recurrent neural networks are proposed for grasping force optimization of multi-fingered robotic hands. The neural networks are shown to be capable of optimizing the norm of grasping force subject to the friction cone constraint and balancing the external force applied to an object. A three-finger example is discussed to demonstrate the optimality of the neural network models.