Abstract
The robust tracking control and accurate real-time uncertainty estimation performances by using disturbance observer-based model reference adaptive control (DOBMRAC) are proposed in this article. An example of coefficient of friction estimation system is studied to demonstrate the proposed method. First, the uncertainty adaptation law in the traditional model reference adaptive control (MRAC) was usually assumed that the uncertainty is a constant or slowly variable. The adaptation law of MRAC could not deal with the time-varying uncertainties. Therefore, a time-varying uncertainty adaptation law by disturbance observer (DOB) is designed to accurately estimate the time-varying uncertainty. Then, a Lyapunov candidate function is implemented to integrate DOB and MRAC as DOBMRAC to perform robust tracking control and accurate real-time uncertainty estimation, simultaneously. The coefficients of friction estimation system are performed to verify robust tracking control and accurate real-time estimation performance by using DOBMRAC. From the simulation results, the proposed DOBMRAC and MRAC are compared, and the proposed method shows well robust tracking control performance and real-time estimation ability. The proposed DOBMRAC has the advantages of rapid convergence responses, robust tracking control, and accurate real-time uncertainty estimation, simultaneously.
Real-Time Uncertainty Estimation in Computer Vision via Uncertainty-Aware Distribution Distillation
