Quantitative performance guaranteed neural adaptive cooperative tracking control of dual-arm robots

Abstract

To achieve high-performance control of dual-arm robots, it is essential to fully consider the overshoot, peak value, settling time, and accuracy of the tracking error. Up to now, the control performance of dual-arm robots has only realized two aspects: settling time and motion accuracy, without considering the overshoot and peak of the transient response. This paper investigates the problem of neural adaptive prescribed performance coordinated tracking control for dual-arm robots, subject to completely unknown robot and object dynamics. By using a fixed-time prescribed performance function and a shifting function, it ensures the quantifiable adjustment of overshoot, peak value, settling time, and accuracy, and guarantees the natural satisfaction of initial conditions. The unknown dynamics of the robot and object are approximated and compensated using a neural network. The stability of the dual-arm robots system and the boundedness of all internal signals are ensured by the Lyapunov method. Additionally, the internal force is ensured to be bounded and capable of minimizing the error to an arbitrary small value. Simulation comparisons have verified the effectiveness and superiority of the proposed method.