Recognition of the three-dimensional shape of objects grasped for PESA multi-fingered robot hand

摘要

Original multi-fingered robot hand can use tactile sensors to recognize the contact state and the force of contact, but is not able to recognize the shape of the object grasped just by tactile without vision. This paper proposed a new recognition algorithm of object grasped applied in PESA hand to solve this problem. The equipment consists of angle sensors installed in the finger joints, tactile sensors installed on the fingers and distance sensors installed on the palm, which can estimate the shape and size information of objects grasped through this algorithm by multi-point grasping repeatedly. The results of experiment shows that this algorithm is effective and can help robot hand measure the three-dimensional shape features of the objects grasped when implement its function.