Robust state/output feedback linearization of direct drive robot manipulators: a controllability and observability analysis

In this research, a robust feedback linearization technique is analysed for robot manipulators control. A complete first-order Taylor series expansion is used to linearize the robot dynamics which takes into account initial conditions and the Taylor-series remainder. A modified PD control law with Taylor-series compensation is used to guarantee robust reference tracking. Whilst classic feedback linearization controllers guarantee asymptotic convergence to zero, the proposed approach shows that, for real applications, if the linearized robot dynamics is stable then the nonlinear robot states are also stable and remain bounded. This premise is assessed via Lyapunov stability theory under a controllability and observability analysis; and hence, exponential convergence to a bounded set is concluded. Experiments are carried out using a 1-degree of freedom robot and a 4-degree of freedom exoskeleton robot to validate the proposed approach.