Internal impedance control helps information gain in embodied perception

摘要

Internal impedance is one of the key factors determining the quality of embodied perception and action in biological organisms and robots. Though the role of impedance control in robotic actuation has been well studied, its significance in the accuracy of proprioception with embodied sensors is not well known yet. Therefore, it is important to characterize the relationship between the entropy of sensor information and the impedance of their physical embodiment, through which sensors feel the internal state of the body and the environment. In this paper, we address the role of internal impedance in the accuracy of embodied perception. To investigate this, we pose the problem of using only torque data measured at the stationary base of a two link planar manipulator, to estimate the deflection caused by an external torque in the McKibben type pivot joint with variable stiffness. Based on analytical modelling and experimental validation, this paper presents, for the first time, that non-linear static memory primitives relating internal impedance, internal kinematic variables, and forces felt at the base of the manipulator - similar to the functionality of tendon organs of biological counterparts - can be used to tune optimal internal impedance parameters to maximize the accuracy of internal state estimation during external perturbations.