Robotic manipulation with large time delay on visual feedback systems

摘要

A difference between the human and robots from the viewpoint of motor control is a sampling time in everyday movements. That is, the sampling time for controlling conventional robot systems is required to become approximately 1 ms. On the other hand, the motor control based on human nervous system permits considerably large time-delay due to the transmission latency on afferent/efferent pathways of the central nervous system. To date, it has been difficult for the robot to acquire dexterous tasks and precise movements as long as the robot system has an unexpected large delay in terms of sensory informations. Based on the above observation, this paper provides a new control strategy to accomplish precise orientation control of a target object grasped by a robotic hand consisting of two degrees-of-freedom (DOFs). The controller, named as serial two-phased (STP) controller, proposed in this study can realize secure grasping and manipulation in the case that a large visual feedback delay induced by the low specification of a camera is hiding in the control loop. Finally, through several simulations, we indicate that the closed loop dynamics designed by integrating the STP controller and a soft-fingertip structure is robust even in at most 100 ms-delay relating to the updating of camera images.