Adaptive controller design for flexible joint manipulators (Q1911265)

This paper presents a new adaptive control algorithm for robotic manipulators with flexible joints, obtained within the singular perturbation approach. The algorithm is based on a decomposition of the robot dynamics into a slow and a fast subsystem, deriving corresponding steady state and boundary layer systems and stabilizing them exponentially. More specifically, the boundary layer system has been stabilized by a high gain PD feedback from the actuator position error, whereas the exponential stability of the steady state system has been achieved by a Slotine-Li type adaptive controller exploiting in its adaptation loop a suitably filtered elastic force signal. Persistent excitation conditions are introduced in order to prove uniform boundedness of the adaptation gain matrix. A simulation study showing the performance of the new controller has been included.