Indirect Adaptive Robust Image-Based Visual Servoing with Precise Online Calibration of Extrinsic Parameters

This paper presents an indirect adaptive robust image-based visual servoing (IBVS) method that improves extrinsic parameter estimate accuracy. Unlike traditional adaptation methods, the proposed approach separates the control law from the adaptation law using indirect adaptation, enabling more efficient parameter estimates. Parameter estimates with nonlinear regression in visual servoing are addressed by the over-parameterization technique. Additionally, the estimated extrinsic parameters are corrected at each step to the closest element of the special orthogonal group SO(3). The correction ensures the physical validity of the parameter estimates, and decreases the necessary persistent excitation (PE) order for fewer parameters to be estimated. This combination of techniques not only improves the accuracy of online extrinsic parameter estimates but also ensures asymptotic trajectory tracking in the pixel frame. Simulation results on an IBVS system with one marker demonstrate the effectiveness of the proposed framework.