Adaptive control for a class of direct drive robot manipulators

Abstract

Direct drive robot technology has received considerable attention by the robotics research community given its potential for developing robots with superior positioning accuracy and repeatability characteristics. Direct drive robots do not have the compliance and backlash problems typically associated with geared robots. With the absence of gearing, however, the motors in direct drive robots must be able to produce considerably larger torques than the motors in a geared robot. As a result, the inductances of the motor windings are much larger, thereby making the electrical motor dynamics no longer negligible. In addition, many motors that are ideal for use in direct drive robots are highly non-linear and consequently their dynamics must be included in the overall control design if the benefits of direct drive technologies are to be fully realized. In this paper we design adaptive controllers for a class of direct drive robot manipulators whose actuators consist of Brushed Direct Current (BDC), and Brushless Direct Current (BLDC) and switched reluctance (SR) motors. In particular, the controllers presented handle the multilink dynamics of a rigid link robot and the multiple-input non-linear electrical dynamics of BLDC and SR motors. The result is a class of adaptive controllers that achieve global asymptotic convergence of the link position tracking error in spite of parametric uncertainty throughout the entire electromechanical robot model. In addition, the controllers only require measurements of link position, link velocity and electrical currents.