Abstract:

Using a multi-fingered robot hand to grasp a target is an important objective in the research of robotics. In most factories or the average family’s household, targets may be located far from each other, so the ability to grasp the target accurately in the approximate position becomes important. Compared with mechanical grippers, which only have one degree of freedom, a multi-fingered robot hand has more operation space to grasp with a higher tolerance of error. This thesis primarily focuses on using a position-based impedance control and virtual spring between robot fingers to change the original trajectory of the robot arm and hand to grasp a target based on the feedback force/torque, which can be read from a tactile array sensor on the robot hand and six-axis force/torque sensors on the wrist when contact is made with the target. The base control, which is a robust adaptive control, can track the trajectory along with the upper control, which is the impedance control, to change the trajectory to grasp the target. A grasping model was constructed to investigate these abilities.