A Damping-based Robust Nonlinear Observer on $SE(3)$

Abstract

This paper presents an improved nonlinear observer structure for a nonlinear system on Special Euclidean Group SE(3) with visual sensing assumed in the presence of modeling uncertainties and measuring noise. In particular, an extra damping term driven by the error between the local observer output and the sensor output is integrated to compensate the robotic system. It is found that this improved observer structure yields a larger domain of attraction and the robust performance is also enhanced, with a carefully selected tuning gain of the damping term. More precisely, it is proved that the input-to-state-stability with a desired domain of attraction can be achieved with the selected damping gain for uniformly bounded modeling uncertainties and measurement noise. Simulation results are provided to support the theoretical findings.