Predictive Active Disturbance Rejection Control of Pan-tilt Visual Tracking System with Parameter Optimization

To improve the tracking performance of the pan-tilt visual tracking system, a Predictive Active Disturbance Rejection Control (P-ADRC) with parameter optimization is proposed. In this approach, the impact of internal parameter uncertainties and external disturbances on the closed-loop system performance is suppressed. Advance prediction of the target dynamics is achieved by Singular Value Decomposition-Unscented Kalman Filter (SVD-UKF), which reduces response lag caused by Tracking Differentiator (TD) in ADRC and visual processing. To realize optimal performance of P-ADRC, the Adaptive Particle Swarm Optimization (APSO) is used to overcome the problems of numerous parameters and difficulty in tuning. Simulation results indicate that, realizing satisfactory transient response and anti-disturbance behavior, the proposed method drastically reduces the Root Mean Square (RMS) tracking error of the pan-tilt system compared to conventional ADRC.