Han Xue

Balancing the rapid control response and the prevention of actuator fatigue, particularly in scenarios involving a large dynamic range, poses a significant challenge. To mitigate frequent actuator start-stop cycles or acceleration-deceleration while ensuring control performance and safety, a nonsingular practical predefined-time control of manipulators with input quantization and a disturbance observer is proposed. A practical predefined-time disturbance observer is developed, which can deal with disturbances with unknown bounds. Subsequently, by adopting logarithmic input quantization, a balance is achieved between rapid control response and the preservation of input data characteristics, particularly in scenarios with a large dynamic range. The density of the logarithmic input quantization is finite. It ensures the accuracy and efficiency of input quantization by preserving the characteristics of the input data, especially in cases with a large dynamic range. Finally, the proposed method is verified using multi-joint manipulators and demonstrates good performance. It enhances convergence speed and disturbance rejection capabilities compared with predefined-time control utilizing linear input quantization.

How to maintain performance and avoid instability when encountering velocity-free measurement and actuator failures in dynamically changing environments is challenging. Through employing an embedded function, a new fixed/predefined-time criteria is introduced. Due to more adjustable parameters for the size of compact set and settling time, the scheme is more general for facilitating the control design, and can converge to a smaller residual set. It can achieve a reduction in the overall trajectory tracking error with the presence of disturbance and uncertainties. Subsequently, a practical fixed/predefined-time fault-tolerant tracking control of manipulators without velocity measurement and under actuator faults is proposed Based on a state observer, it achieves predefined-time estimation of joint angular velocity using only the signals from position sensors. A predefined-time observer is used for estimate the upper bound of lumped disturbance, and another one is used to compensate for actuator faults and saturation. Finally, the proposed control method is verified through trajectory tracking of manipulators.

Input delay has a significant impact on the stability and performance of formation control systems. However, the issue of practically predefined‐time convergence in the presence of such delays is seldom addressed in the literature. To enhance control performance in heterogeneous systems with different dimensions or orders under input delays, this study introduces a novel practically predefined‐time control algorithm and singularity avoidance for scenarios requiring urgent control responses. To the best of our knowledge, this is the first time to propose an input delay tolerance of practically predefined‐time formation control for heterogeneous multiple unmanned vehicles with mixed dimensions and mixed degrees of freedoms. The proposed algorithm demonstrates adherence to stability conditions for second‐order heterogeneous multi‐agent systems. Moreover, the settling time can be preset by users. Sufficient conditions for achieving predefined time convergence are identified and verified through unmanned surface ships and unmanned underwater vehicles in trajectory tracking scenarios. The boundedness of the control system is confirmed using the Lyapunov theorem, addressing the challenge of real‐world scenarios with control speed requirements. This approach offers significant improvements in the safety and efficiency of transportation under input delay.

How to achieve predefined-time stability in the presence of both actuator faults and Denial-of-Service attacks presents a significant challenge. Herein, an aperiodic intermittent predefined-time tracking control method is proposed to deal with threat scenarios in the presence of actuator faults. An intermittent predefined-time stability criterion is introduced. The control law is alternated among three operational modes: the working interval, resting interval, and holding interval. During an attack on the actuator channel, the input is temporarily set to zero. Conversely, if an attack targets the sensor channel, the control input retains the value from the last successful transmission. The proposed method is validated through trajectory tracking of second-order systems, such as manipulators. The results demonstrate that this approach reduces reliance on communication bandwidth and decreasing the risk of exploitation from potential attacks.

With the presence of external disturbance and moving targets, how to establish new predefined time stability criteria to achieve prescribed performance under varying sea states is a challenging problem. A new predefined‐time stability lemma is introduced through employing an exponentially decaying function. It has more adjustable parameters for the size of residual set, and can enter a smaller residual set. Subsequently, according to this lemma, a practical predefined‐time prescribed performance control for cooperative target enclosing of unmanned surface vehicles is proposed. A predefined‐time observer is designed to compensate for uncertain parameters and disturbances. Finally, experimental results validate that this control scheme reduces tracking errors when compared to existing methods, and demonstrates robustness against external disturbances.

Given that time-consuming adjustment processes can result in system instability or substantial performance degradation, it is crucial to ensure stability within a predefined timeframe without requiring prior knowledge of the attack signals. A practical predefined-time reinforcement learning control mechanism is designed. To the best of our knowledge, this is the first time to propose the reinforcement learning with predefined-time convergence subjected to actuator attacks. Subsequently, a theoretical proof of the proposed control algorithm is provided. The flexibility of the proposed algorithm is reflected in its use of predefined-time reinforcement learning to handle actuator attacks, measurement limitation, and unmodeled errors, while employing Barrier Lyapunov Function to ensure that the state tracking errors meet the constraint conditions. The effectiveness of this approach is verified through its application to two-wheeled robots for trajectory tracking. Consequently, the settling time for both position and angle can be decreased by 77.8% compared to reinforcement learning and by 66.7% in relation to fixed-time reinforcement learning control method. Ultimately, this research enhances the transient stability and robustness of two-wheeled robot control under actuator attacks and disturbance.

The presence of time delays prevents the controller from responding to disturbances promptly, resulting in excessive overshoot and prolonged adjustment times, which in turn deteriorate system stability. To address the issue, a practical predefined-time control approach is proposed for systems with state time delays. The Lyapunov–Krasovskii functional method is employed, without the use of trigonometric or exponential functions. Additionally, a practical predefined-time observer is developed to estimate mismatched disturbances, enhancing the robustness of nonlinear time-delay systems. A practical predefined-time neural network is also used to estimate system uncertainty. The proposed approach is validated through experiments. The impact of delays on the system is analyzed under multiple scenarios. Compared to traditional control methods, the results show a reduction in computational costs and improved control performance in the presence of time delays. The proposed method outperforms the conventional predefined-time control algorithm by 7.5% and shows an improvement of approximately 20% over the fixed-time control method.