Nonlinear model reference adaptive control using tap-delay filters

A model reference adaptive control (MRAC) technique realized with tap delay (TD) filters is presented. For the purposes of exposition, the technique is applied to the control of a simple two-link, planar robot with revolute joints. The adaptive logic is based on Lyapunov stability. The TD filter allows a straightforward implementation of an inverse modeling control scheme that can be used in the initiation of motion, before the MRAC system can effectively compensate. The resulting control system is, computationally, very simple and requires adaptation of as few as five weights per joint and measurement of only link angular orientation and rate. The control scheme is evaluated in simulation. The effects of disturbance torques and large payload variations are examined, and it is demonstrated that the system yields excellent performance in a series of large-amplitude step-response tasks