Disturbance suppression has long been a challenging issue in the study of adaptive dynamic programming (ADP) algorithms. This paper introduces a novel critic neural network (NN)-based robust ADP algorithm for a class of continuous-time affine nonlinear systems to handle the unknown bounded external disturbances. The algorithm adopts an improved robust cost function (RCF), which simultaneously enhances the capability of rejecting the disturbances and reduces the controller's energy consumption. Based on this, a disturbance compensator is further proposed to achieve asymptotic stability of the closed-loop system. To facilitate the application of the proposed control algorithm to practical engineering problems, a critic-only NN is employed to approximate the optimal cost function using a novel tuning law. Consequently, the proposed critic NN and RCF-based optimal control algorithm, combined with the RCF-based disturbance compensator and NN tuning law, ensures the uniform ultimate boundedness (UUB) of the system's state trajectory. The effectiveness of our algorithm is demonstrated through software simulations.

This paper investigates the predefined-time tracking control for Quadrotor Unmanned Aerial Vehicles (QUAVs) subject to external disturbances, actuator faults, and attitude state constraints. To overcome the singularity issues and conservative parameter constraints inherent in conventional predefined-time control, a generalized predefined-time convergence theorem is proposed by introducing an improper integral method, which relaxes the Lyapunov derivative constraints and greatly improves design flexibility. On this basis, a nonsingular adaptive backstepping control scheme is developed. Specifically, an improved piecewise barrier function is embedded to transform constrained states into an unconstrained domain, achieving a better trade-off between constraint compliance and tracking precision. To compensate for the lumped disturbances, including external disturbances and actuator faults, an adaptive predefined-time disturbance observer is designed without prior boundary information. Meanwhile, a differentiator is introduced into the controller to suppress the "explosion of complexity" problem, and the adaptive compensation term is constructed to mitigate the estimation errors of the observer and filter. Lyapunov stability analysis proves that all closed-loop signals are uniformly ultimately bounded, and that the tracking errors converge to a small vicinity of the origin within a user-defined time. Finally, numerical simulations and real-world flight experiments are conducted to verify the high accuracy, strong robustness, and practical applicability of the proposed scheme.

We present an online mini-batch GPR-powered forwarding controller for nonlinear systems. The design learns lumped uncertainties in real time and injects data-driven compensation into the virtual control of a reduced-order target system, using GP predictive variance for uncertainty-aware robustification. We derive explicit upper bounds on the minibatch GPR prediction error and provide input-to-state stability of the closed-loop system with respect to the learning error. On a flexible robot-arm case study, the method achieves fast model convergence and tighter tracking with lower variance than nearest-neighbor and RBF-NN backstepping/forwarding baselines. The results provide a deployable route to calibrated probabilistic learning in real-time feedback control.

Ground-based simulation of low-/micro-gravity environments is an important enabling technology for space exploration, and suspended gravity offloading (SGO) systems have been widely employed because of their practicality and scalability. However, the two-dimensional tracking subsystem of an SGO system is subject to strong dynamic coupling, underactuated characteristics, external disturbances, and model uncertainties, which makes high-precision tracking control difficult to achieve. To address these challenges, this paper proposes an adaptive predefined-time disturbance observer-based sliding mode control framework for the two-dimensional tracking subsystem of SGO systems. First, an adaptive predefined-time disturbance observer is developed to estimate the lumped disturbance caused by external disturbances and model uncertainties. Then, a nonsingular predefined-time sliding mode controller is designed to improve tracking accuracy and robustness while avoiding the singularity problem commonly encountered in terminal sliding-mode designs. Compared with conventional finite-time and fixed-time control methods, the proposed method allows the upper bound of the convergence time to be explicitly prescribed by a user-defined time parameter and is independent of the initial conditions. Finally, comprehensive numerical simulations are conducted to verify the tracking performance, disturbance rejection capability, and robustness of the proposed control strategy.

This paper investigates the fixed-time tracking control problem of an unmanned aerial vehicle (UAV) considering the disturbance, input saturation, and actuator failure. According to the hierarchical control principle, the UAV dynamics are decomposed into a translational and rotational loop to accommodate the controller design. A novel nonsingular fixed-time backstepping controller based on switching variables is proposed to achieve fast convergence of system tracking errors within a fixed time. To overcome the effect of the disturbance and the actuator failure, two fixed-time disturbance observers are designed in two loops, respectively. By integrating the fixed-time auxiliary variables into the dynamic controllers, the problem of input saturation can be addressed. In addition, the tracking errors of the closed-loop system converge to the neighborhood of the origin in a fixed time. Finally, sufficient simulation results verify the validity of the proposed control framework for the UAV.

This paper studies secure forwarding control for a class of nonlinear systems where the measurements are subject to passive attacks. To make sure the security of measurements, a chaotic encryption algorithm is developed to mask the plain measurements in the transmitter. Based on system invertibility, a chaotic decryption algorithm is proposed to recover the plain measurements in the receiver. As for reference tracking, the input-to-state stability from the decryption error to state is theoretically guaranteed. Specifically, the decryption error is attenuated by a robust forwarding controller and the tracking error is ensured to converge to a neighborhood around zero. We conduct simulations to demonstrate encryption and tracking performances through a case study on flexible joint robots.

This paper presents a new method to achieve approximate-optimal control of second-order nonaffine nonlinear systems with unknown dynamics, external disturbance, and input saturation by integrating a pre-compensator, the adaptive dynamic programming (ADP) technique, and a disturbance compensator. The pre-compensator transforms the nonaffine system into an affine-like one and integrates a saturation condition into the augmented system to address input saturation. A new quadratic cost function based on sliding mode (SM) is proposed for the augmented system to simplify the cost function's dimensionality, which enhances the efficiency of the algorithm. A compensation term is further proposed to counteract the unknown bounded disturbances. We will show that, the designed robust sliding mode-based ADP (SMADP) guarantees the asymptotic stability of the closed-loop system and, by utilizing neural networks (NNs) to approximate the optimal value of the cost function, the resulting NN-based SMADP algorithm can also guarantees the uniform ultimate boundedness of the trajectory of the closed-loop system. The simulation results validate the effectiveness of our SM-based robust ADP control method.

This paper introduces a novel robust cost function for the ADP algorithm to suppress the impact of the disturbance for partially unknown second-order nonlinear systems. The sliding mode control is employed to design the cost function, which thereby enhances the performance of the closed-loop system. Furthermore, an offline disturbance compensator is proposed to further improve the robustness of the system. Subsequently, the trajectory of the entire closed-loop system is proved to be uniformly ultimate bounded (UUB) in a Lyapunov sense. Sufficient simulations are conducted to verify the effectiveness and superiority of the proposed method.

Tackling the prevalent challenge of unknown model elements and perturbations in practical systems poses a significant barrier to enhancing control precision. This paper proposes a novel RBF-based Nonlinear MPC for mode compensation. Initially, the conventional approach of dynamic modeling is utilized to identify and isolate unmodeled characteristics. Subsequently, Radial Basis Function (RBF) neural networks are employed to predict and compensate for these unmodeled parts. Driven by the sampled data, this method efficiently explores the control action space to improve control performance. Our three-layer neural network architecture significantly reduces computational overhead, and online network updates effectively mitigate neural network generalization issues. We apply the proposed approach to force tracking control of Antagonistic Pneumatic Artificial Muscles (APAM) in flexible structures. Case studies demonstrate a significant improvement in control accuracy compared to the feedforward PID control method.

In this work, an attractive alternative of robust backstepping approach for a reference tracking problem is developed. We extend robust forwarding design strategy to a class of nonlinear systems with unknown functions. In particular, using radial basis function neural networks (RBFNNs), uncertainties and derivatives of mappings in the recursive computation are estimated with any arbitrarily small approximation error. Two salient features include: first, nonlinear model is not required to be linearly parametrizable; second, at each step, the problem of “explosion of terms” is tackled by applying neural networks instead of filters. The boundedness of all signals is proven and the tracking error is ensured to converge to a small neighborhood of zero. Simulation results and comparisons with robust backstepping are given to demonstrate tracking performances through a case study on flexible joint robots.

Parameter optimization is a crucial area within the field of control theory. This study introduces a novel framework based on reinforcement learning (RL) for controlling quadrotors. Initially, fast nonsingular terminal sliding mode control (FNTSMC) serves as the fundamental trajectory tracking controller for the quadrotor. Subsequently, fixed-time disturbance observers (FTDO) are employed to mitigate disturbances. Ultimately, an RL training framework is introduced to optimize the hyperparameters within the FNTSMCs. Extensive simulation and physical experiments are conducted to validate the efficacy and superiority of the proposed control framework.

To address the challenge of target tracking for non-maneuvering,single-station setups in long-range scenarios,we propose a target tracking algorithm leveraging three-dimensional angle of arrival data,characterized by its asymptotically unbiased nature.Initially,we construct a motion and observation model centered on a non-maneuvering single station,assuming a known rate prior,and examine the system's ob-servability.To tackle the bias inherent in the pseudo linear least squares algorithm,we introduce a con-strained total least squares method that demonstrates asymptotically unbiased properties,with its effective-ness validated through simulations.In tests involving three-dimensional angle tracking over distances in the hundred-kilometer range,with angle measurement standard deviations at 0.1°,0.2°,and 0.3°,the con-strained total least squares method achieves a time-average relative distance error of 6%,12%,and 21%within 50-100 seconds,respectively,and an absolute position error of 9 km,19 km,and 35 km;at initial distances of 70,140,and 280 km,the errors are 1%,6%,and 30%for the same duration,with absolute errors of 0.7 km,9 km,and 30 km.Notably,the relative distance error can be reduced to below 10%within 100 seconds,marking a significant precision enhancement,while maintaining operational speed comparable to the pseudo linear least squares method.The constrained total least squares approach exhibits rapid convergence,high accuracy,and swift processing,showing resilience against angle measurement er-rors and initial distance variations.It offers a robust solution for 3D angle of arrival tracking of non-maneu-vering single-station targets in distant settings.

To address the issue of underactuation caused by passive buffers in a tether-driven microgravity simulation system,an active buffer control method based on Pneumatic Artificial Muscles(PAM)was proposed.Firstly,the tether-driven microgravity simulation system was analyzed for ground-based micro/low-gravity simulation.A block-structured nonlinear neural network modeling method to effectively over-come the highly nonlinear nature of PAM was introduced.Subsequently,the disturbances caused by the flexible interaction between the tether and the spacecraft was analyzed.Finally,a nonlinear model predic-tive tracking control approach was employed.Compared to traditional PID control methods,the proposed was introduced approach offers advantages such as simple parameter adjustment,excellent real-time track-ing performance,and robust control performance in the presence of perturbations to the target inertia pa-rameters of the unloading system.Experimental results demonstrate that the proposed method ensures tracking force error within 3%under various disturbances.The feasibility of the active buffer based on PAMs is confirmed experimentally,and the proposed control method achieves force-tracking control in the presence of flexural uncertainty.

To address the underactuation issue induced by passive dampers in a suspension Gravity offload(SGO) system, this paper introduces the utilization of active control using Pneumatic Artificial Muscles(PAM) to transform it into a fully actuated physical system. However, due to the inherent non-linear characteristics such as flexibility and hysteresis in PAM, achieving precise force control poses challenges. Therefore, this paper proposes a Neural Network-based Nonlinear Model Predictive Control (NMPC) approach. We apply the proposed approach to the constant force control of the SGO system based on PAM. Simulation results demonstrate a marked improvement in control accuracy when compared to the feedforward PID control method.

To intercept the maneuvering target at a desired terminal angle, this paper presents a time-varying sliding mode guidance law with consideration of the second-order autopilot dynamics and input saturation. To achieve the finite-time interception and satisfactory overload characteristics, a time-varying sliding mode guidance law is developed, which enables the line-of-sight (LOS) angle error to converge into a small neighborhood of the origin at the interception time. An auxiliary system is constructed to reduce the adverse effect generated from the input saturation. Moreover, with the aid of extended state observers, the proposed guidance law requires no information on the target acceleration and the acceleration derivative of the interceptor. The performance of this guidance law is verified via the numerical simulations.

The problem of a missile attacking maneuvering targets has been considered in this paper.Meanwhile, a new integrated guidance and control (IGC) method for interceptors with the impact angle is constrained has been proposed. Aimed at achieving the convergence in finite-time ,as well as satisfactory input characteristics, this paper presents a time-varying sliding mode control (TVSMC) method, in which a time-base generator function (TBG) is introduced. Then, this IGC scheme is developed based on the TVSMC and back-stepping technique. At each step, the derivative of the virtual signal is approximated by a tracking differentiator (TD). To enhance the system robustness, the fractional power extended state observer (FPESO) is introduced,for estimating the lumped disturbances and the target maneuvering. Based on the Lyapunov tools, the closed-loop system states are proved to converge into tiny regions around the origin state at the interception time. Finally, the IGC law proposed in this paper has been proved to be effective.

This paper proposes a multi-stage dueling deep Q-network (MS-DDQN) algorithm to address the high-speed aerial vehicle evasion problem. High-speed aerial vehicle pursuit and evasion are an ongoing game attracting significant research attention in the field of autonomous aerial vehicle decision making. However, traditional maneuvering methods are usually not applicable in high-speed scenarios. Independent of the aerial vehicle model, the implemented MS-DDQN-based method searches for an approximate optimal maneuvering policy by iteratively interacting with the environment. Furthermore, the multi-stage learning mechanism was introduced to improve the training data quality. Simulation experiments were conducted to compare the proposed method with several typical evasion maneuvering policies and to reveal the effectiveness and robustness of the proposed MS-DDQN algorithm.

This paper proposes a finite-time command filtered backstepping guidance law (FCFBGL) with the terminal angle constraint while accounting for the input saturation and the autopilot dynamics. To eliminate the adverse effect induced by the filtering errors and the acceleration saturation, a new finite-time error compensation mechanism is integrated in the guidance law design. The proposed FCFBGL not only guarantees the the line-of-sight (LOS) angle error to converge to a small neighborhood of the origin in finite time but also achieves the continuity of the input signal. in finite time. Moreover, with the aid of the fractional power extended state observer (FPESO), the proposed FCFBGL requires no information on the target acceleration and the acceleration derivative of the missile, which is preferable in the practical application. The finite-time stability of the proposed guidance law is derived with the Lyapunov methodology. Simulation results illustrate the effectiveness and superiority of the proposed guidance law.

This paper deals with the problem of intercepting maneuvering targets with terminal angle constraints for missiles subjected to three-dimensional non-decoupling engagement geometry. To achieve the finite-time interception and satisfactory overload characteristics, a time varying sliding mode control methodology is developed based on a time base generator function. The main feature of the proposed guidance law guarantees the Line-of-Sight (LOS) angles to converge to small neighborhoods of the desired values at the interception time. First, a fractional power extended state observer is used to estimate the unknown target acceleration, which can significantly reduce the adaptive switching gain. The fractional power extended state observer enjoys the advantage of better noise tolerance. Then, a newly designed sliding mode surface is constructed by introducing a time base generator function and the time-varying sliding mode guidance law is developed based on this time-varying sliding surface. The proposed guidance law significantly reduces the overload magnitudes. Numerical simulations are carried out to verify the performance of the proposed guidance law.

Considering the problem of a skid-to-turn (STT) interceptor attacking maneuvering targets in the three-dimensional space, a new integrated guidance and control (IGC) law with the constraints of impact angles and input saturation is developed. To achieve the finite-time convergence and satisfactory input characteristics, a time-varying sliding mode control (TVSMC) method is presented based on a time base generator function. To this end, the IGC scheme is developed based on the TVSMC and the back-stepping. At each step, the derivative of the virtual signal is approximated by a tracking differentiator. Meanwhile, an auxiliary compensation system is designed to reduce the adverse effect raised from the saturation error. Moreover, the modeling uncertainties and disturbances are attenuated effectively by using fractional power extended state observers to estimate them. The closed-loop system states are proved to be uniformly ultimately bounded and the controlled states are proved to converge into small neighborhoods of the origin at the interception time. Finally, the performance of the presented IGC law is verified through numerical simulations.