An Adaptive Grasping Strategy for Dexterous Hands Based on Proximity-Contact Sensing

In this paper, we propose a dexterous hand grasping strategy based on proximity sensing sensors and contact sensors. The strategy does not need make any assumptions about the properties and surface features of objects. This sensing system consists of two parts: a pose detector to construct the surface near dexterous hand and a slip signal detector to calculate the slip amplitude. These two sensing components are integrated to form a new sensor PT-Tip. When grasping an unknown object, the dexterous hand uses proximity sensors to sense the surrounding environment and find a safe grasping point, then the contact sensor is used to measure the grasping contact process. The covariance matrix of feedback information from tactile sensor is calculated to determine whether the object slips or not. The strategy for three-finger grasping experiments is verified on two objects with different shapes.