A globally convergent observer for velocity estimation in robotic manipulators with uncertain dynamics

We present a method for global estimation of joint velocities in robot manipulators. A non-minimal model of a robotic manipulator is used to design an adaptive observer capable of handling uncertainties in robot dynamics. Dimension of the proposed observer is shown to be at least 3n where n stands for the manipulator degrees of freedom. This number is less than the dimension of most of existing globally convergent adaptive observers. Global asymptotic convergence of system state estimates to their true values is achieved under no persistency of excitation condition. Smoothness of the dynamics of the proposed observer allows its easy implementation in comparison with non-smooth observers. Simulation results illustrate low noise sensitivity of the proposed observer in comparison with non-smooth observers.