Three-Finger Grasping Optimization for Push-Button Objects

Grasping problems have been commonly encountered in practical robotic applications. Existing researches usually focus on seeking optimal grasp points such that a rigid object can be kept in hand without falling down. Such methods may fail when the object is deformable and needs to be further manipulated while being grasped. This paper focuses on three-finger grasp planning for push-button type objects. The main challenge is that the grasp needs to keep stable during the deformation of the manipulated object. By analyzing the mass distribution variation during the manipulation, we establish an optimization problem for inducing a resilient grasp that can adapt to the push-button object grasping and manipulation scenario. The experiments show that an optimal grasp for a pure rigid object grasp problem may fail to maintain a stable grasp when the push-button object is manipulated, while the grasp provided by our method is always resilient to the manipulation.