Keywords: Robotics, Motion Control, Optimal Control
Abstract: This study addresses the challenge of generating a contiguous sweeping line along an irregular curb edge. The primary method involves utilizing the curb edge line and applying a translation process to it. Subsequently, a higher-order filter is employed to smooth the resulting curve. The curve is then segmented based on curvature, with spiral and Reeds-Shepp curves used to connect the segmented parts. Finally, the entire trajectory is smoothed using the Augmented Lagrangian-iLQR (AL-iLQR) algorithm, yielding the final curve for the sweeping line. This approach ensures a precise and optimized path for cleaning along the irregular curb edge.

Keywords: Robotics, Motion Control, Real-time Systems
Abstract: The development of physically assistive robots for eating assistance has significant potential to improve the quality of life for seniors and individuals with disabilities. Effective autonomous feeding, however, depends on accurately inferring eating intentions. Many existing methods fail to adequately address human intention and the dynamic variability of behavior during eating. This paper presents a method for inferring eating intentions in human-robot interaction (HRI) by combining Residual Networks (ResNet) with Long Short-Term Memory (LSTM) networks, enabling robotic arms to feed autonomously. Inspired by observable facial movements during eating, we extract a key feature: the amplitude of the chin-to-nose distance from facial landmarks, which clearly indicates eating intentions. To mitigate noise and data corruption in input sequences, we use a Gaussian function as the convolutional kernel in the ResNet framework and integrate a variance attention mechanism in the LSTM’s hidden layer to capture dynamic changes. Experimental results show our method achieves an accuracy rate 87.0% in intention inference. Real-world tests with a robotic arm and an RGB camera validate our approach's efficacy and real-time predictive performance.

Keywords: Robotics, Motion Control, Real-time Systems
Abstract: This paper introduces a novel evaluation network that leverages deep reinforcement learning integrated with advanced modeling and prediction strategies to enhance decision-making in racket sports, drawing parallels to challenges encountered in unmanned autonomous systems (UAS). Traditional performance analysis methods predominantly depend on manual observation and static metrics, which inadequately capturethe dynamic and strategic complexities of game environments. Our approach combines the principles of Markov Decision Processes with the Transformer architecture to manage long sequential tasks, thereby improving the accuracy of correlating states and actions. By modeling turn-based racket sports, the evaluation network assigns Q-values to actions derived from historical match data, demonstrating the impact of each action on potential scoring or loss of points. We evaluated our approach using various hyper-parameters and network architectures, validating it against multiple baselines and performance metrics. The findings suggest promising avenues for advancing data-driven training methodologies and enhancing autonomous system capabilities in complex, task-oriented domains.

Keywords: Robotics, Control Applications, Motion Control
Abstract: This paper explores a quadrotor pursuit-evasion (PE) game in an environment with obstacles. Existing research typically focuses on evasion strategies that avoid obstacles while increasing the distance from the pursuer, and very few studies have attempted to use obstacles to gain an advantage in winning the PE game. Assuming both quadrotors follow Dubin's car dynamics and the pursuer is less agile due to a larger minimum turning radius, this paper introduces a novel evasion strategy. This strategy leverages the obstacle by having the evader perform agile turns near it, potentially causing the pursuer to collide with the obstacle. Gazebo simulations have been conducted to validate that the evasion strategy can lead to the pursuer colliding with the obstacle under appropriate conditions.

Keywords: Robotics, Nonlinear Systems and Control, Control Applications
Abstract: This paper presents a trajectory tracking control approach for a mobile manipulation (MM) comprising a robotic arm mounted on a mobile platform. To address external disturbances and non-inertial interactions between the two subsystems, a composite trajectory tracking control strategy based on finite-time prescribed performance control (FTPPC) is developed. This method ensures precise trajectory tracking despite rapidly varying and bounded uncertainties. An adaptive sliding mode observer (ASMO) is designed to estimate external disturbances and uncertain system dynamics. Using a Lyapunov-based analysis, the proposed scheme guarantees uniform boundedness and uniform ultimate boundedness of the system. Simulation results validate its effectiveness in achieving high-precision tracking and superior transient performance under uncertain conditions.

Keywords: Robotics, Nonlinear Systems and Control, Modeling and Control of Complex Systems
Abstract: Quadrotors frequently encounter sudden external forces during flight, often resulting in loss of control and crashes. Conventional control algorithms typically fail to effectively handle these large external disturbances. To address this issue, we propose a novel control framework that integrates Model Predictive Control (MPC) with Active Disturbance Rejection Control (ADRC), alongside a bidirectional thrust geometric control algorithm. When the external force remains within a predefined threshold, the quadrotor utilizes the disturbance rejection capabilities of ADRC to maintain stability. Conversely, if the force exceeds this threshold, the quadrotor leverages the external force to execute a complete 180-degree flip to regain equilibrium. Simulation results are presented to demonstrate the superior performance of the proposed control framework in handling large external disturbances.

Keywords: Learning-based Control, Linear Systems
Abstract: The data-driven output regulation problem via internal model principle has been studied by both policy-iteration method and value-iteration method. But the results were limited to single-input single-output linear systems with zero input-output transmission matrix. Recently, we extended the existing results to multi-input multi-output linear systems with non-zero input-output transmission matrix by an improved value-iteration method. Since the policy-iteration method is simpler and has a much faster convergence speed than the value-iteration method, in this paper, we further establish the results parallel to the improved value-iteration method by an improved policy-iteration method. Compared with the existing policy-iteration results, we are able to handle multi-input multi-output linear systems with non-zero input-output transmission matrix. Moreover, we further improve the existing policy-iteration algorithm by significantly reducing the computational cost and weakening the solvability conditions. A numerical example is used to illustrate the advantages of the improved algorithm.

Keywords: Multi-agent Systems, Networked Control, Nonlinear Systems and Control
Abstract: In this paper, we study the distributed Nash equilibrium (NE) seeking problem for aggregative games with players whose actions are governed by high-order integrator dynamics over jointly connected and weight-balanced switching networks. Since the existing approaches critically relied on the every time connected network assumption, they do not apply to jointly connected switching networks, which can be disconnected at any time instant. To deal with the jointly connected and weight-balanced switching networks, we propose an approach that is quite different from the existing approaches, which involves finding a time-varying quadratic Lyapunov function for the closed-loop system by using converse Lyapunov theorem. A numerical example regarding formation control of unmanned aerial vehicles (UAVs) is used to validate our result.

Keywords: Learning-based Control, Automated Guided Vehicles
Abstract: In recent years, with the rapid development of unmanned aerial vehicle (UAV), aerial refueling has garnered increasing attention. During the critical docking phase of aerial refueling, traditional sensing technologies such as GPS and electro-optical systems may fail when signals are obstructed, drawing attention to vision-based solutions. Conventional image-based visual servo (IBVS) control techniques often overlook the impact of disturbances. In this paper, an iterative-learning-based IBVS (ILB-IBVS) controller is proposed, extending traditional position-based iterative learning to the 2D image plane, making the control more precise and reliable. Simulation results show that the proposed ILB-IBVS controller improves docking success rates and efficiency compared to the original IBVS controller.

Keywords: Learning-based Control, Robotics, Control Applications
Abstract: In this paper, an active disturbance rejection controller (ADRC) parameter tuning method for a fixed-wing unmanned aerial vehicle (UAV) is proposed. First, a six-degree-of-freedom nonlinear model of a fixed-wing UAV is established, and the attitude controller of the UAV is built based on ADRC. Then, the Pigeon-Inspired Optimization (PIO) is enhanced based on Directional crossover (DC) and Directional variation (DV), and DXPIO is proposed to improve the searching and convergence ability of PIO. Finally, the cost function of ADRC is designed based on sigmoid function for parameter optimization training of DXPIO. In the experiments, the benchmark optimization performance of DXPIO has a significant advantage over the other 7 peers in terms of both search and development capabilities. Additionally, DXPIO is used to optimize the UAV pitch-roll controller separately, and the tuned ADRC controller is compared to the traditional proportional-integral-derivative (PID) controller when gust interference is added on the UAV's body axis. The results demonstrate that the adjusted ADRC controller has improved robustness, anti-interference rejection, and reaction time.

Keywords: Learning-based Control, Intelligent and AI Based Control, Motion Control
Abstract: This paper proposes an autonomous driving framework for unsignalized intersections. The framework consists of two modules: vehicle state inference and vehicle motion planning. A network based on Graph Convolutional Neural Network (GCN) and Gated Recurrent Unit (GRU) encoderdecoder is established. Relying on the relevant historical states of vehicles within the unsignalized intersection, the interactions among vehicles are fully taken into account, and the future trajectory distributions of all vehicles and the collision risks between each pair of vehicles are predicted together. The motion planning module, based on the prediction results of future states, adopts the Model Predictive Control (MPC) method to obtain the desired acceleration input, improving traffic efficiency while focusing on safety. We have built a simulation platform for an unsignalized intersection that simulates human interactions in the Sumo simulation software, and used it for dataset generation and simulation experiments. The results show that our framework demonstrates high efficiency and safety under different traffic volumes.

Keywords: Learning-based Control, Robust and H infinity Control, Nonlinear Systems and Control
Abstract: This study aims at the robust iterative learning model predictive control (ILMPC) for uncertain nonlinear systems with time delays. In order to achieve the H infinity tracking performance, a novel robust ILMPC scheme is designed based on the two-dimentional (2-D) system theory. To deal with the unknown time delays, an appropriate Lyapunov-Krasovskii functional is constructed, and the stability conditions are derived using linear matrix inequalities (LMIs). Furthermore, a new control gain determination strategy is introduced to improve the efficiency of the proposed ILMPC method. The effectiveness is verified through numerical simulations.

Keywords: Robotics
Abstract: In this research, we propose a mechanism that strengthens the robot itself by incorporating surrounding objects in the natural world. The strength and durability required when using a robot vary greatly depending on the application. Therefore, it is necessary to have a mechanism that can switch its own rigidity according to the surrounding conditions. Many of the stiffness-changing robots reported in the past use the phase transition of materials. Although this approach is useful, but there are not many variable stiffnesses. In this research, we examined two types of robot mechanisms that can take in surrounding objects and change their own rigidity. Operational experiments were performed to evaluate the usefulness of each mechanism. We also conducted stress measurements to confirm how much the stiffness actually changes.

Keywords: Robotics, Learning-based Control, Control Applications
Abstract: The automation of electric vehicle (EV) charging is a critical challenge in robotics, requiring high precision and adaptability to handle the complex geometry of charging ports. Traditional methods, such as visual servoing and force control, often struggle to achieve reliable performance due to their inability to simultaneously address positional and force-related uncertainties. To address these limitations, this paper proposes a novel coarse-to-fine framework that integrates visual and force feedback for robust and efficient peg-in-hole operations. Our method leverages a structured light camera for 6-DoF (Degrees of Freedom) pose estimation and an end-mounted force sensor for real-time force feedback. In the coarse localization stage, a learning-based object detection model identifies the charging port, while template alignment refines the pose estimation. In the fine assembly stage, a force-position hybrid control strategy ensures precise alignment and insertion. Extensive real-world experiments demonstrate the effectiveness of our approach, achieving a 96% success rate in inserting the charging gun into EV charging ports.

Keywords: Robotics, Man-machine Interactions
Abstract: This paper explores how path width and pedestrian density affect human-robot interaction in outdoor and retail environments. A mobile robot, equipped with a depth camera and 2D LIDAR sensor is operated on pavements of varying path widths and in retail areas with different pedestrian densities in Gera-Lusan, Germany. Unobtrusive field measures ensured the collection of raw interaction data, which was analyzed by grouping it based on path width and comparing it across different environment conditions. The findings highlight the complexity of human-robot proximity and show that path width can not be the only factor defining the spatial relationships in robot-human interaction. Parameters such as pedestrian density, environment of operation (outdoor and retail) and predefined rules for entry and exit in retail spaces affect the spatial relationship dramatically. Unconsidered factors such as age, gender, personal experiences, weather, robot speed, time of day, and weekday vs. weekend effects may further shape interactions. By challenging simplified assumptions, this study emphasizes the need for a more nuanced approach to design robots for real-world environments.

Keywords: Robotics, Learning Systems, Signal Processing
Abstract: 3D human pose estimation (3DHPE) from a single RGB image is a crucial task in computer vision. The estimation accuracy and speed of 3DHPE directly affect the practical applicability. However, existing methods often improve accuracy by using complex network architectures or multi-stage processing. These approaches result in more computational overhead and slower estimation speed. To balance estimation accuracy and speed, we need a more efficient approach. To address this issue, we propose an end-to-end model called I-KDnet, which achieves high estimation accuracy with fast estimation speed. Specifically, we design an Idealized Knowledge Distillation (IKD) training approach, a idealized variant of online knowledge distillation. During training, I-KD enhance the training process, similarly to online knowledge distillation. During inference, it does not introduce any additional computational overhead. Additionally, compared to online knowledge distillation, the I-KD approach is easier to implement and more effective. Based on this approach, I-KDNet sets a new benchmark for single-frame monocular 3DHPE, achieving best accuracy on the Human3.6M dataset with fast inference speed.

Keywords: Robotics, Optimal Control, Control Applications
Abstract: This paper proposes a model predictive interaction control (MPIC) framework based on impedance matching, embedding impedance regulation into the predictive optimization process. The proposed approach ensures seamless transitions between impedance control in unconstrained situations and optimal control adaptation under task-specific and physical constraints, enhancing interaction safety, robustness, and adaptability. A unified robot-environment interaction model is formulated by incorporating series-parallel interaction properties to simultaneously consider the impact of external perturbations and robot reference position variations on force prediction and optimization. Simulation and experimental results validate the effectiveness of MPIC over conventional impedance control (IC) in terms of constraint handling, disturbance rejection, and balancing compliance and precision, providing a scalable and adaptable solution for complex robot-environment interaction.

Keywords: Energy Efficiency, Estimation and Identification
Abstract: In demand response (DR), customer baseline load plays an important role in determining economic compensation. Traditionally, it is believed that private data cannot be used when estimating baseline loads. This paper proposes a privacy protection scheme based on blockchain to make the use of private data possible. Based on this scheme, the baseline load can be estimated by combining customers private data and non-private data, in which the convolutional neural network (CNN) is employed to capture the relationships of these data and estimate baseline load. This method can be applied to accurately determine the authenticity of customers demand response and solve the problem of baseline load estimation for continuously participating in demand response. Finally, a comparison between our method and other estimation method demonstrates that our study can improve the rationality and reliability of baseline load estimation.

Keywords: Automated Guided Vehicles, Robotics
Abstract: In modern warfare, unmanned aerial vehicle (UAV) swarms frequently encounter signal interference or denial from adversaries, posing significant challenges to localization accuracy. Nevertheless, the large-scale advantage of UAV swarms enables interfered UAVs to utilize relative navigation techniques by leveraging communication and relative measurements. Thus, we discuss a multi-agent distributed cooperative localization method based on ultra-wideband (UWB) modules. First, a distributed Kalman filtering-based multi-UAV cooperative localization algorithm is proposed, achieving high-precision localization under partial GNSS-denied conditions. Second, a multi-agent distributed cooperative localization algorithm incorporating an analysis of UWB non-stationary measurement noise characteristics is introduced, enhancing the robustness of swarm navigation in dynamic formations. Finally, a UWB based distributed multi-UAV cooperative localization prototype system is developed, comprising five small UAVs equipped with inertial measurement unit (IMU), GPS modules, and UWB modules. Experimental evaluations under various scenarios compare and analyze swarm navigation performance under different conditions.

Keywords: Learning-based Control, Multi-agent Systems, Intelligent and AI Based Control
Abstract: Since deep neural networks (DNNs) are inherently black-box models, providing formal stability and performance guarantees for DNN-based controllers remains a challenge. In this paper, we address the formation control problem of multi-agent Lagrangian systems and propose a distributed structured deep neural network (DSNN) controller with inherent stability guarantees. By leveraging the backstepping technique and carefully designing the neural network structures, our controller ensures stability for any set of neural network parameters. Additionally, we derive an explicit upper bound on the formation error in the presence of disturbances, which can be adjusted by tuning the neural network parameters. The effectiveness of the proposed controller is validated through multiple simulations.

Keywords: Multi-agent Systems, Learning Systems
Abstract: This paper considers solving distributed stochastic optimization problems with smooth, nonconvex objective functions over peer-to-peer networks with non-i.i.d. datasets. While existing studies suggest that gradient tracking methods can mitigate the impact of data heterogeneity among nodes, our analysis shows that in nonconvex cases, it can degrade convergence performance due to the consensus error in the initialization of gradient tracking, which inherently reflects non-negligible data heterogeneity. To address this issue, we propose an improved warm-up distributed stochastic gradient tracking algorithm, termed W-DSGT, and theoretically show that reducing the consensus error of the gradient tracking during the initial phase can effectively alleviate the impact of data heterogeneity, leading to enhanced convergence performance.

Keywords: Networked Control, Multi-agent Systems, Control of Distributed Generation Systems
Abstract: In this paper, we propose a novel consensus algorithm for multiagent networks (MANs) operating under deception attacks and privacy disclosures. First, to address the exposure of real-time agent state information during transmission, a time-varying weighted iteration mechanism is developed based on the PushSum algorithm, ensuring secure protection of agent privacy. Second, to mitigate the impact of external attackers on consensus, the weighted iteration mechanism is integrated with the W-MSR algorithm, enabling secure system convergence. Finally, through mathematical theory analysis proved that the algorithm can effectively protect the privacy of agent initial state information and ensure the MAN to achieve resilient consensus under the deception attacks.

Keywords: Robotics, Control Applications, Motion Control
Abstract: This paper explores the deployment of a mobile unmanned aerial vehicle carrier (UAVC) system, facilitating the landing of unmanned aerial vehicles (UAVs) on a moving platform, thereby enhancing their operational range and flexibility. The primary contributions of this study include the development of an advanced trajectory planner that integrates Jump Point Search (JPS) with gradient-based trajectory optimization to ensure efficient and collision-free navigation in complex environments. Furthermore, a Nonlinear Model Predictive Control (NMPC) framework is employed to achieve precise and stable trajectory tracking for both the UAV and UAVC. Extensive simulations conducted in Gazebo validate the efficacy of the proposed approach, demonstrating successful landings on a UAV carrier under a complex environment.

Keywords: Nonlinear Systems and Control, Robotics, Adaptive Control
Abstract: The payload position control problem for the quadrotor with a cable-suspended system is challenging due to its underactuated nature, particularly when the actuation dynamics are considered. To address these limitations, we transform the payload system into a fully actuated system by modeling the quadrotor as an equivalent actuator. We then design a hierarchical control framework based on the backstepping technique, which consists of the payload position controller, the swing angle controller, and the adaptive actuator controller to control both the payload position and quadrotor actuation. The asymptotic stability of the overall system is proved by the Lyapunov method. Finally, the feasibility of the proposed methodology is demonstrated through simulations.

Keywords: Nonlinear Systems and Control, Optimal Control, Robotics
Abstract: Aiming at the issues of touchdown overshoot or low-altitude floating caused by mismatches in the landing trajectory angle or sink rate during UAV spot landing on sloped runways, which may affect landing accuracy and safety, this paper proposes a landing trajectory optimization and tracking control strategy adapted to sloped terrain. First, a segmented optimization method for spot landing trajectory adapted to sloped runway terrain is proposed based on the Gauss Pseudospectral Method (GPM). Then, a position and attitude decoupled flight control method for spot landing on a sloped runway is proposed based on the direct lift control concept. Additionally, controllers are designed using Dynamic Surface Control (DSC) technology, along with Extended State Observer (ESO) to estimate system states and disturbances during the landing process. Simulation results validate the feasibility and effectiveness of the proposed landing strategy.

Keywords: Nonlinear Systems and Control, Robotics
Abstract: Controlling a cooperative transportation system transporting a cable-suspended payload is challenging due to highly coupled dynamics and unpredictable disturbances. In this paper, a novel time-varying uncertainty and disturbance estimator (TV-UDE) is introduced to estimate and compensate for disturbances in real time dynamically. Unlike conventional fixed-gain UDE observers, the proposed TV-UDE method smoothly adjusts its gain over time: starting large to mitigate transient spikes and increasing to a small gain for accurate steady-state disturbance rejection. This strategy enhances robustness and stability by avoiding the peaking phenomenon associated with high-gain estimators. The asymptotic stability of the overall system is proved by the Lyapunov method. Finally, the feasibility of the proposed methodology is demonstrated through simulations.

Keywords: Motion Control, Learning-based Control, Nonlinear Systems and Control
Abstract: In complex map environments, the Rapidly-exploring Random Tree (RRT) algorithm is recognized as an efficient initial path planning method for unmanned aerial vehicles (UAVs). Relying solely on centralized computation or the planning capabilities of a single node often faces challenges such as limited device resources, vulnerability of navigation points to interception, and insufficient real-time adaptability. This paper proposes a distributed path optimization framework based on federated learning (FL) and edge computing (EC), referred to as DPOFEC, which formulates the path optimization problem as a global optimization task within the framework of FL. First, edge nodes of the server execute initial path planning using the RRT algorithm to generate local path segments. Then, the FL framework aggregates the weights uploaded by each edge node and optimizes the path points. Finally, the enclosed and safe sphere-shaped corridors are designed around the optimized global path points, with the size of these corridors dynamically adjusted to accommodate obstacle distributions and the UAV's flight state. Experiments demonstrate that in a simple scenario (Case 1), the proposed method improves path generation processing time by approximately 38% and 43%, compared to traditional RRT algorithm attempts 1 and 2, respectively. In a complex scenario (Case 2), the improvements are approximately 51% and 40%, respectively. Leveraging distributed collaboration, the algorithm enhances the performance and robustness of path planning while effectively protecting the privacy of path point data.

Keywords: Linear Systems, Control Applications, Motion Control
Abstract: Electric Vertical Takeoff and Landing (eVTOL) aircrafts have garnered considerable interest as a fundamental component of urban air mobility. However, the saturation is likely to happen due to the large mass and rotational inertia, especially when motor failure occurs in the eVTOL system. Therefore, this article tries to analyze the characteristics of saturation performance of control systems, particularly the second-order integral system, which is the most common system in trajectory following and control. Firstly, the overshoot condition of the saturation system is derived by phase portrait analysis. Then, the relation of the overshoot ratio with the command input, control bandwidth, and input constraints is given. After analyzing the saturation characteristics, the conclusion is applied to design the eVTOL controller for enhancing control performance and satisfying airworthiness.

Keywords: Learning-based Control, Control of Distributed Generation Systems, Control of Smart Power Delivery Systems
Abstract: Abstract— Traditional unit commitment (UC) does not consider distribution grid constraints, while common transmission-distribution coordination methods require substantial computational costs. As the distributed energy resources (DERs) in distribution grids continue to grow, traditional distribution grids are gradually transforming into active distribution grids (ADGs), when solving UC, it is necessary to consider the constraints of the distribution grid to ensure the operational safety of the transmission and distribution grids. This paper proposes a UC approach with ADGs by using learning-based power flow constraints substitution which can accelerate the computation and promote the integration of DERs while ensuring the safe operation of the distribution grid. We trained two multi-layer perceptrons (MLPs) and converted them into binary variable linear constraints equivalent to substitute for power flow constraints. The experiment confirms that the suggested strategy performs better in power scheduling when based on the IEEE118-bus system as the transmission grid, and ten IEEE33-bus systems as distribution girds.

Keywords: Learning-based Control, Learning Systems, Optimal Control
Abstract: In this paper, a model-free off-policy constrained adaptive dynamic programming (CADP) algorithm is developed for discrete-time general nonlinear systems. By using an adaptive learning rate, the controller update is constrained in a well-defined neighborhood to avoid performance degradation in case of high convergence uncertainty of the CADP mechanism. The learning mechanism is used to tune a Proportional-Integrative-Derivative (PID) controller parameters in a model-free offline fashion. The Markovian assumption is proved for the closed-loop system with a general PID controller to show its applicability in Adaptive Dynamic Programming (ADP) like algorithms. The validation performed on a nonlinear process highlighting the stability-preserving attributes of the proposed method using the constrain mechanism.

Keywords: Learning-based Control, Optimal Control, Learning Systems
Abstract: Reinforcement Learning Model Predictive Control (RL-MPC) has achieved significant progress in recent years. However, existing approaches still have some limitations. This paper proposes a Bayesian policy exploration method for RL-MPC that substantially enhances its performance. Specifically, we implement Bayesian posterior estimation of value functions and introduce an optimistic exploration strategy tailored for efficient exploration of RL-MPC, which improves the sample efficiency of RL policy exploration. Then an optimistic Bayesian exploration strategy is proposed, which encourages the agent to leverage existing model information to achieve superior control performance. The soundness and effectiveness of our method are evaluated through an empirical study of controlling a drone to reach targets subject to uncertain model parameters and environmental perturbations. The results validate that our approach has superior performance compared with benchmarks.

Keywords: Learning-based Control, Learning Systems, Nonlinear Systems and Control
Abstract: A method utilizing reinforcement learning and transfer learning for hexacopter attitude control is proposed. The reinforcement learning approach employs Proximal Policy Optimization (PPO), which enables continuous output and prevents abrupt policy updates, ensuring stable learning. Transfer learning, which accelerates learning by adapting a previously trained agent to a similar environment, is applied to develop a fast and efficient hexacopter attitude controller. The research process consists of three major phases. In the first phase, a quadcopter attitude control policy is trained using reinforcement learning to achieve stable flight and maneuvering. In the second phase, the output of the pre-trained quadcopter control model is modified to facilitate its transfer to a hexacopter. In the final phase, the pre-trained model is used as the initial learning model, and transfer learning is applied to extend the quadcopter controller to a hexacopter attitude control system. The performance of the transfer-learned hexacopter controller is validated through dynamic simulations. To evaluate the effectiveness of transfer learning, a comparative study is conducted by measuring the time required for a hexacopter controller to reach full training completion using standard PPO and comparing it to the training time when transfer learning is applied. Instead of a single measurement, multiple training sessions are performed, and the average required training time is analyzed. The results demonstrate that transfer learning enables the hexacopter to achieve similar performance in a shorter amount of time compared to training from scratch using PPO.

Keywords: Learning-based Control, Motion Control, Intelligent and AI Based Control
Abstract: Efficient path planning is crucial for autonomous UAV navigation in dynamic environments. Traditional 3D path planning methods rely on precise environmental models while simultaneously facing challenges in handling high-dimensional complexity and discontinuous speed control. To overcome these challenges, we propose a UAV path planning approach for 3D obstacle spaces based on the Proximal Policy Optimization (PPO) algorithm. Our method integrates a recurrent neural network into the PPO framework to process time-series data effectively, enabling the UAV to leverage historical information for improved decision-making in dynamic scenarios. The exploration is further enhanced by designing an environmental reward scheme that combines angle-based rewards with a curiosity module, which helps the UAV rapidly discover optimal paths. To ensure a smooth flight trajectory, the planned path is optimized using the Minimum Snap method. The overall approach is validated on the UE4+AirSim platform demonstrating its practical applicability for real-world UAV flight decision-making. The experimental results demonstrate the superiority of this method.

Keywords: Smart Buildings, Smart Structures
Abstract: Evaluation of facade degradation is crucial to determining the need for further examination and maintenance, safeguarding the structural health of the building. Traditional evaluation methods rely on visual inspection and subjective judgment by surveyors, which requires working at heights and manual detection and severity assessment. This process is time-consuming, poses safety risks, and may result in potential errors. To address these challenges, unmanned vehicles have been deployed for building inspection tasks, saving time and minimizing safety concerns. Additionally, diverse deep learning techniques have been utilized to automate the visual evaluation process and reduce subjective errors to classify, detect, and segment defects. However, previous research has primarily focused on improving image processing accuracy without incorporating the entire evaluation process into industry evaluation standards. In this paper, we propose a framework for automated building facade defect evaluation that can be applied to both unmanned and manned data collection systems. Our approach employs the deep neural network for defect segmentation. The trained model accurately recognizes individual defects and extracts their properties, such as width, length, and area using morphological operations. Following ISO standards, evaluation results are automatically obtained within defined effective evaluation areas. Furthermore, defect information and evaluation results are registered to a model generated with 3D reconstruction techniques, providing a valuable reference for experts in formulating maintenance plans. Finally, experiments are conducted on a tall building to verify the effectiveness of our proposed method.

Keywords: Fault Detection and Diagnostics, Signal Processing, Estimation and Identification
Abstract: Ice build-up on wind turbine blades is a significant issue, leading to operational risks and reduced efficiency. Traditional detection methods, such as visual inspection, power curve analysis or specialized sensors, are often slow, inefficient, or costly. This paper proposes an approach using 0-dimensional persistence homology from topological data analysis (TDA) on tower and blade vibration spectra to extract key features representing the lifespan of the sub-level sets to formulate a clearer supervised learning problem. Persistence diagrams are embedded in persistence images and rank functions, enabling ice detection through convolutional neural networks (CNN), and functional principal component analysis (FPCA) with support vector machines (SVM). This approach shows promise for reducing equipment and maintenance costs, leading to more efficient blade monitoring and maintenance processes.

Keywords: Fault Detection and Diagnostics, Control Applications
Abstract: This paper presents a deep transfer learning-based approach for diagnosing multiplicative faults in underwater manipulators using limited operational data. Given the limited availability of data, transfer learning is utilized to enhance model performance. A pre-trained model from conventional manipulators is adapted to the underwater domain through model-based transfer learning. The convolutional neural networks (CNNs) and long short-term memory (LSTM) networks are used to retain both local and temporal fault characteristics, improving fault feature extraction. The source domain model is fine-tuned using a small sample dataset from the target domain, where lower layers are frozen and the top layers are fine-tuned for fault diagnosis, achieving improved accuracy. The results from the case study demonstrate that the proposed approach delivers high accuracy in diagnosing actuator faults in underwater manipulators.

Keywords: Fault Detection and Diagnostics, Learning Systems, Signal Processing
Abstract: Prognostics and Health Management integrated bearing remaining life prediction not only supports condition-based maintenance decision closed loop, moreover, the on-time task rate is improved by dynamic scheduling of maintenance resources, and the continuous improvement of equipment comprehensive efficiency is promoted. In this paper, a hybrid-driven Remaining Useful Life (RUL) evaluation way of bearing is constructed, which innovatively realizes the deep coupling of data-driven and mechanism model by fusing improved DenseNet and nonlinear least square degradation modeling. First of all, the time-domain statistical features of vibration signals are selected by using monotonicity and robustness, and the degradation trajectory index(DI) is established to determine the Fault Incipience Point (FIP). Secondly, a joint feature learning framework of Continuous Wavelet Transform and Causal Dilated Convolution (CWT-CDC) was constructed: the time-frequency map was generated by Morlet wavelet kernel function, and the local damage feature was enhanced by causal dilated convolution, and the nonlinear mapping from equipment Health Indicators (HI) to RUL value was realized. Finally, the superiority of the proposed appoach are verified by the XJTU-SY bearing data set.

Keywords: Fault Detection and Diagnostics, Smart Buildings
Abstract: Abstract—Digital Twin (DT) introduces a novel paradigm of autonomous virtual-physical integration for fault diagnosis in Environmental Control Systems (ECS), with its core lying in the self-adaptive optimization of diagnostic capabilities through dynamic data mapping and logical evolution. Despite the progress made by existing DT-driven fault diagnosis methods in hybrid modeling, rule mining, and adaptive analysis, significant challenges persist. These challenges include the effective fusion of heterogeneous sensor data and the evolution of diagnostic logic under dynamic conditions. This paper systematically reviews core technical pathways for fault diagnosis, analyzes theoretical limitations based on current research, and proposes an innovative Digital Twin-Fault Tree Analysis-Association Rule Mining (DT-FTA-ARM) collaborative diagnostic frame- work. The proposed framework employs advanced algorithms to balance its computational efficiency with interpretability, addressing critical issues in multi-source data alignment and cross-system knowledge transfer. Field validation in subway ECS applications demonstrates the framework's potential to enhance operational reliability while adhering to strict latency and energy constraints. Index Terms—Digital Twin, Fault Diagnosis, Environmental Control Systems, Hybrid Modeling

Keywords: Fault Detection and Diagnostics, Sensor/Data Fusion, Intelligent and AI Based Control
Abstract: The collaborative analysis of both cyber-layer and physical-layer data is crucial for improving detection accuracy and timeliness of cyber-attack. Cyber-layer features provide early indicators of attacks, while physical-layer features reflect the actual impact on the power system. To leverage this synergy, a cross-attention mechanism is introduced to generate cross-layer features to capture these cross-layer interactions. Furthermore, based on the traditional Mixture of Experts (MoE), a novel framework MoE-Transformer Dual Layers Detection (MoE-TransDLD) is proposed, which dynamically fuses multi-layer features to model cyber-physical dependencies. Specially, MoE-TransDLD assigns a dedicated expert to each layer, including a cyber-layer expert, a physical-layer expert, and a cross-layer expert, to more accurately model multi-layer data relationships in power systems. Notably, both the expert network and the gating network share a common Transformer architecture to extract global features, while maintaining corresponding independent feed-forward network (FFN), where each expert focuses on its respective domain and the gating network achieves adaptive and dynamic selection in decision making. The synthetic Texas 2000-bus model system is used as an experimental model and its physical-layer data and cyber-layer data are collected. The experimental results show that the MoE-TransDLD significantly outperforms the existing methods and achieves superior classification metrics and faster attack detection time.

Keywords: Fault Detection and Diagnostics, Real-time Systems, Intelligent and AI Based Control
Abstract: Supervisory Control and Data Acquisition (SCADA) signals such as wind speed and power output are frequently used for fault detection in wind turbine (WT) condition monitoring. However, there has been limited attention on SCADA alarm analysis, despite its importance in managing turbine reliability. This paper presents an automated methodology to analyze SCADA alarm signals for identifying patterns predictive of turbine faults and minimizing downtime. To minimize downtime and identify turbine fault patterns, this paper proposes an automated methodology for analyzing SCADA alarm signals. First, alarms are classified into normal, abnormal, and questionable events using a structured taxonomy. These categories are linked to instances of downtime in a later analysis. Alarms are ranked according to their operational impact and criticality using a prioritization method that was modified from industry standards. The results show that even low-volume alarm data holds significant diagnostic and prognostic value. The suggested event-and pattern-based alarm analysis improves the ability to identify faults and provides proactive maintenance planning. This work underscores the value of intelligent alarm handling to reduce downtime and improve operational reliability in wind farms.

Keywords: Intelligent and AI Based Control, Learning-based Control
Abstract: This study addresses the limited autonomous target-acquisition capabilities exhibited by unmanned wingman systems in beyond-visual-range (BVR) combat scenarios. We propose a hierarchical reinforcement learning framework based on Proximal Policy Optimization (PPO). The dual-layer framework integrates six-degree-of-freedom (6-DoF) flight dynamics with radar lock-on control systems, where the low-level controller achieves precise tracking of attitude parameters (heading, airspeed, altitude) with steady-state errors below 1%. Extensive simulations conducted on the JSBSim platform show an 89% success rate in target acquisition across three combat scenarios: random maneuver engagement, head-on interception, and target pursuit. The proposed geometric mean reward formulation with roll angle constraints effectively reduces stall risk. This framework provides an interpretable decision-making architecture for cooperative UAV operations in modern aerial combat systems, overcoming dynamic adaptability constraints in conventional expert systems.

Keywords: Modeling and Control of Complex Systems, Sensor/Data Fusion
Abstract: In this paper, the problem of rapidly reconstructing the kill chain for tasks due to the frequent changes of battlefield situation in the current beyond-visual-range cooperative air combat is studied. The kill chain is approximated as a series of action combinations arranged in chronological order for the target. Based on the idea of sequential decision-making, the reconstruction problem is transformed into a task redistribution problem at each stage of the kill chain. When the reconstruction condition is triggered, the full-chain reconstruction is performed, which reduces the complexity of the problem. By using the improved particle swarm optimization algorithm and relying on the kill chain effectiveness evaluation as the objective function, the combat resources and combat tasks are screened and matched. Through simulation experiments, the results obtained by the algorithm are compared with the results obtained by the enumeration method, which verifies that the algorithm can improve the speed of problem solving while ensuring the effectiveness of the results.

Keywords: Intelligent and AI Based Control, Learning Systems, Learning-based Control
Abstract: The rapid advancement of artificial intelligence has driven the application of Deep Reinforcement Learning in multi-agent systems, offering new opportunities for complex collaborative scenarios in transportation, energy, and communications. As urban traffic congestion intensifies, adaptive traffic signal control based on Multi-Agent Reinforcement Learning (MARL) has emerged as an effective approach to optimize traffic flow and reduce intersection delays. However, existing research largely focuses on single-intersection control, with limited exploration of coordinated multi-intersection scheduling. This study proposes a MARL-based method for coordinated traffic signal optimization, utilizing Multi-Agent Proximal Policy Optimization (MAPPO), which demonstrates superior stability and convergence over other algorithms. To address MARL's dimensionality challenges, we introduce two key innovations: (1) optimizing the MAPPO network architecture by integrating a Transformer module to enhance model expressiveness and accelerate convergence; and (2) incorporating Large Language Models (LLMs) to leverage their reasoning capabilities for improved multi-agent collaboration. Experimental results show significant improvements in traffic control efficiency, laying a solid foundation for intelligent transportation systems.

Keywords: Control Applications, Optimal Control, Smart Buildings
Abstract: The following paper examines Data-Enabled Predictive Control (DeePC) for combined temperature and humidity control in a historical museum building. Bypassing classical offline system identification techniques, DeePC uses a nonparametric model only based on previously collected closed loop input-output data. Within a co-simulation consisting of MATLAB and EnergyPlus, the control performance while ensuring the preventive conservation requirements is evaluated. The impact of both the hyperparameters and measurement noise is investigated. The results show that DeePC maintains temperature and humidity within the limits for cultural asset protection. Integrating disturbances and weather forecasts into DeePC to enhance control performance and energy efficiency could be considered in future work.

Keywords: Optimal Control, Adaptive Control, Learning-based Control
Abstract: Quadrotor's agile flight in complex environments has numerous potential applications such as search and rescue. Recently, nonlinear model predictive control (NMPC) has shown more advantageous results in agile quadrotor control. However, it relies on highly accurate models for maximum tracking accuracy and lacks the capability to reject external disturbances. Model uncertainties, including unmodeled complex aerodynamic effects and external disturbances, will degrade the system's performance. In this paper, we propose gaussian process-L1-nonlinear model predictive control (GP-L1-NMPC), a novel hybrid adaptive NMPC approach that leverages gaussian process regression to learn complex unmodeled aerodynamic effects and employs L1 adaptive control to compensate for external disturbances in real time. Specifically, we use the nominal model enhanced with the gaussian process model as a reference model for the L1 adaptive control to reduce tracking error. The proposed method demonstrates immense tracking accuracy and robustness, with more than 90% tracking error reduction over baseline NMPC without any gain tuning at a speed of 10 m/s.

Keywords: Optimal Control, Multi-agent Systems, Modeling and Control of Complex Systems
Abstract: This paper proposes a pursuit-centric Game-Theoretical MPC (GT-MPC) framework for quadrotors, leveraging Nash equilibrium to anticipate evader maneuvers and optimize pursuit strategies. Unlike conventional MPC or reinforcement learning (RL) methods, GT-MPC explicitly models bidirectional adversarial interactions, enabling 63.03% faster average capture times in symmetric pursuit scenarios. By focusing on strategic parity—where pursuer and evader share identical dynamics—we demonstrate that superior decision-making, not hardware advantages, can also drive capture efficiency. Simulations across 50 randomized trials validate GT-MPC’s robustness, with a 93% success rate under perfect information, outperforming state-of-the-art baseline.

Keywords: Signal Processing, Control Applications
Abstract: Sound carries rich information, yet its utilization in drone-based applications, particularly in extracting environmental information from ego-noise, has received limited attention. This paper researches the interaction between drone ego-noise and its corresponding echoes, exploiting phase cancellation effects to detect obstacles. A waveform detection algorithm has been developed, integrating the Root Mean Square (RMS) of sound energy to enhance obstacle detection rates. The performance indicates that it is feasible to use acoustic information in autonomous robotics for obstacle detection, especially in high-noise mobile platforms such as drones.Future research will focus on increasing detection accuracy and enabling precise obstacle distance estimation.

Keywords: Fault Detection and Diagnostics, Estimation and Identification
Abstract: This paper proposes a fault detection and isolation method to deal with actuator fault in four-wheel steering (4WS) vehicles. First, a lateral dynamic model for 4WS vehicles is developed by incorporating the T-S fuzzy model, while accounting for external disturbance and sensor measurement noise. This improved model is designed to address the unmodeled dynamics caused by variation in longitudinal velocity. Then, a novel residual generation observer based on the H_-and L_∞ performance indices is designed, aiming to balance the fault sensitivity and disturbance robustness. Furthermore, the concept of directional residuals is introduced, and the isolation of faulty actuator is achieved by analyzing the directional correlations between the fault feature vectors and the generated residual vectors. Finally, the effectiveness of the proposed method is validated through simulation experiments.

Keywords: Estimation and Identification, Modeling and Control of Complex Systems
Abstract: Distributed driving electric vehicles (DDEV) has become one of the research hotspots in automobile industry due to its fully decoupled chassis configuration. Accurate and timely estimation of road adhesion coefficient (RAC) is the premise of vehicle active safety, which will greatly improve the driving comfort and handling stability of DDEV. In this research, an adaptive unscented Kalman filter (AUKF) is proposed for RAC estimation to deal with issues of model parameter perturbation and unknown time-varying noise. Fading coefficient matrix is used to enhance the utilization of new observation data. Meanwhile, Sage-Husa noise estimator is adopted to refresh system noise dynamically. Results indicate that AUKF method has a higher precision, faster convergence and better stability than EKF and UKF. The proposed AUKF method also shows strong robustness to different pavement coefficients and has fortissimo generalization ability to multiple driving scenarios.

Keywords: Adaptive Control, Fault Detection and Diagnostics
Abstract: With the continuous advancement of intelligent connected vehicle technology, the presence of vehicle faults and network attacks poses significant threats to the reliability and security of electric vehicle. This paper proposes an adaptive neural network (NN) fault-tolerant controller design for vehicle, and employs the dynamic watermarking method to detect stealthy replay attacks. First, a vehicle dynamics model is established considering actuator faults and unknown disturbances. Then, a fault-tolerant controller is designed by integrating an adaptive NN and a nonlinear disturbance observer (DO) to handle faults and external disturbances. Next, by incorporating dynamic watermarking into the controller, a residual-based detection function is designed to detect stealthy replay attacks. Subsequently, the convergence of the proposed control algorithm is proven using the direct Lyapunov method. Finally, hardware-in-the-loop (HIL) tests validate the effectiveness of the proposed control method.

Keywords: Smart Buildings, Learning Systems, Real-time Systems
Abstract: Current building inspections are mainly relied on human, making the process time-consuming and labor-intensive. Therefore, the development of deep-learning-based automated inspection systems has become a growing trend. However, the lack of a comprehensive training dataset is the majority challenge in developing deep-learning models. To bridge this gap, we introduce CUBIT-CW, a human-annotated dataset containing 10871 high-resolution images collected via drone, with multiscale ranging from 800 x 600 to 8000 x 6000 pixels. Our CUBIT-CW covers 9 distinct defect categories, which are line cracks, map cracks, rust stains, delamination, spalling, cement loss, seepage, damp patches, efflorescence and mold. It also works for instance segmentation, allowing for the models to have a detailed understanding of defect boundaries and features. To evaluate its effectiveness, we test CUBIT-CW by 6 stateof-the-art(SOTA) computer vision models. The corresponding experimental results illustrate that our dataset enhances the robustness and generalizability of the models, making them to inspect more complex defects on building surfaces with greater accuracy.

Keywords: Intelligent and AI Based Control, Learning-based Control, Control Applications
Abstract: In this paper, an autonomous interception method is presented for intelligent high-speed vehicles (HSV) with matrix game theory, incorporating an enhanced pigeon-inspired optimization. Considering attackers with equal levels of intelligence, the matrix game is applied to describe the real-time one-to-one interception process. Subsequently, a maneuver library based on a simplified overload model is introduced to predict the motion tendency of a 6-degree-of-freedom (6-DOF) model. In this regard, individual similarity pigeon-inspired optimization (ISPIO) is proposed to search for optimal maneuvers. The experiments demonstrate the effectiveness of the proposed framework in face-to-face interception and initial disadvantage scenarios, and analyze the influences of different optimization algorithms for both scenarios.

Keywords: Learning Systems, Fuzzy and Neural Systems, Intelligent and AI Based Control
Abstract: An automatic multiple pedestrian detection and counting framework is introduced in this research paper. The proposed technique combines successfully several computer vision and nonlinear anisotropic diffusion-based models. Its detection component performs a boosted cascade classifier-based walking person localization process that is followed and improved by a segmentation of the detected sub-images, which is performed applying a novel partial differential equation (PDE) - based geodesic active contour (GAC) model. The obtained pedestrian detections are then counted successfully in the analyzed traffic video by applying a tracking-by-detection scheme using a deep learning-based feature extraction to them. The detection and tracking simulation results that are finally discussed illustrate the effectiveness of the proposed framework.

Keywords: Intelligent and AI Based Control, Learning-based Control, Nonlinear Systems and Control
Abstract: To improve the fault-tolerant adaptability of hypersonic gliding vehicles (HGVs) in diverse environments and actuator faults, an enhanced intelligent fault-tolerant control (FTC) scheme is proposed based on the deep reinforcement learning (DRL) and transfer learning (TL) method. First, the HGV model with multiple disturbances and diverse actuator faults is constructed. On this basis, a fundamental FTC system is designed under the active disturbance rejection control (ADRC) framework to provide theoretical stability and basic FTC capability. Then, a DRL-based intelligent FTC scheme is proposed to autonomously adapt the FTC parameters to different environments and faults. Furthermore, to address the discrepancies between the actual flight environments and virtual training environments, as well as the issue that the actual faults exceed the knowledge of the FTC agent, a TL-based enhanced intelligent FTC scheme is proposed to ensure that the FTC agent can continuously update its policy using actual flight data, thereby improving its adaptability to more complex and unknown conditions.

Keywords: Intelligent and AI Based Control, Multi-agent Systems, Robotics
Abstract: The problem of UAV attack-defense confrontation is a hot research direction in the field of unmanned systems at present. However, in an environment with threat areas or obstacles, when the enemy is a defender with higher mobility or pursuit ability, the cooperative penetration problem of multiple UAVs is still lack of effective solutions. Therefore, this paper combines the theoretical analysis of game theory and the advantages of reinforcement learning in complex scenes, and designs an algorithm framework for embedding dominant region analysis into deep reinforcement learning. On the premise of sacrificing strategy, we analytically derive the attacker's dominance region through geometric optimization and integrate this framework into the Deep Deterministic Policy Gradient (DDPG) algorithm by enhancing state space formulation, reward function design, and termination criteria. Numerical simulations demonstrate the algorithm’s superior efficacy over baseline reinforcement learning approaches, exhibiting reduced training time (42.8%), increased penetration success rate (31%), and optimized trajectory lengths (4.9%).

Keywords: Intelligent and AI Based Control
Abstract: Fecal Microbiota Transplantation (FMT) has emerged as a promising therapeutic approach for various gastrointestinal and systemic diseases. However, optimizing donor-recipient matching remains a critical challenge that constrains its clinical efficacy. In this study, we propose EnteroMatch, a deep learning-based Sparse Mixture of Experts (Sparse MoE) model designed for FMT donor-recipient matching. By integrating a dynamic routing mechanism, EnteroMatch effectively captures the complex ecological characteristics of the gut microbiota. Furthermore, we employ k-means clustering to partition both donors and recipients into two distinct enterotypes, allowing the model to adaptively adjust to their unique microbial profiles and better reflect the influence of microbiome diversity on FMT outcomes. Extensive experimental evaluations on large-scale datasets demonstrate that EnteroMatch outperforms other state-of-the-art deep learning architectures in terms of matching accuracy, generalization, and robustness. This work not only provides a novel computational framework for personalized FMT strategies but also lays a solid foundation for future research in microbiome-based therapies.

Keywords: Intelligent and AI Based Control, Robotics, Smart Structures
Abstract: This paper introduces an open-source, lightweight, compact projectile launching device designed to fill the gap in target acquisition systems for micro aerial vehicle (MAV) pursuit-evasion and dogfighting research. This device unifies projectile storage, feeding, and launching functions in a single integrated design. When fully loaded with projectiles, it weighs only 70.5 g, making it highly compatible with small MAVs (e.g., 3.5-inch MAVs). The device features a single friction wheel mechanism and a spring-slide rail design for projectile feeding, enabling stable launches in all orientations. Furthermore, we developed a trajectory prediction model by integrating aero- dynamic principles and the Magnus effect, specifically adapted to full-pose launch conditions. Based on the projectile impact dispersion, we proposed a noise model which can enhances the accuracy and portability of the capture. Experimental validation confirms the stability of launching projectiles and the accuracy of the trajectory predictions. This study achieves, for the first time, a hardware-model co-design for full-orientation projectile launching in MAVs, advancing close-range MAV combat technologies towards practical applications. The related results have been made publicly available via GitHub.

Keywords: Multi-agent Systems, Networked Control
Abstract: In the event-triggered control of multi-agent systems (MASs), external disturbances are prevalent in practice and may lead to excessive triggering events (often referred to as Zeno behavior), which poses problems for practical implementation. Therefore, it is essential to enhance event-separation properties when the MAS is subjected to disturbances. In this context, this article is concerned with the event-triggered consensus of MASs in the presence of external disturbances. Distributed dynamic event-triggered (DET) control strategies are proposed based on the sampled information of neighboring agents. It has been theoretically demonstrated that, under the designed DET sampling strategies, the MAS can achieve bounded consensus and strictly positive minimum inter-event times are guaranteed. The effectiveness of the theoretical results is validated by numerical results.

Keywords: Multi-agent Systems
Abstract: This paper deals with the distributed time-varying optimization problem over a digragh (or directed graph). Motivated by the time-varying nature present in the cost functions, we model the time-varying features using an exosystem and formulate the problem of minimizing a global cost function, which is the sum of the local time-varying cost functions. We design a distributed algorithm for each agent that only utilizes the information of its own cost function and the information obtained through a network represented by a strongly connected and weight-balanced digraph. Convergence analysis is carried out to show that the decision variables of all agents converge to the time-varying optimal solution with time. Simulation results verify the theoretical conclusions.

Keywords: Multi-agent Systems, Control Applications, Optimal Control
Abstract: The cooperative localization of multiple drones in space with a certain configuration to target enhances the acquisition of information and boosts situational awareness, presenting significant application prospects. Addressing the challenge of achieving high-precision cooperative localization for maneuver targets, this paper designs a Nash-based game cooperative localization strategy. Firstly, it establishes the cooperative localization scenario as a Stackelberg model and considers both localization accuracy and input constraints. Secondly, it theoretically derives the Nash equilibrium solution of the game model for multi-aircraft. Thirdly, an improved data-driven adaptive dynamic programming algorithm with independent actions is devised to solve the equilibrium solution. Finally, simulations verify that the proposed model and algorithm can achieve cooperative localization of maneuvering targets by multiple drones, meeting the requirements for localization accuracy. This provides a solution for the research of strategies in cooperative adversarial scenarios.

Keywords: Multi-agent Systems, Networked Control
Abstract: This paper addresses the resilient leader-follower consensus problem for discrete-time high-order multi-agent systems with time-varying graphs. A resilient control law is proposed for each normally functioning follower to effectively mitigate the disruptive effects of malicious agents. Under the proposed control law, resilient leader-follower consensus is achieved exponentially, given that the graph is jointly strongly (2f+1)-robust, significantly reducing the communication burden at each time step. Finally, the effectiveness of the proposed approach is validated through a numerical simulation.

Keywords: Multi-agent Systems, Networked Control, Robotics
Abstract: Conflicts often arise in swarm robotics between individual tasks related to environmental adaptation and the cooperative objective of maintaining a formation. For instance, obstacles may prevent robots from achieving a prescribed formation. Individual tasks, such as collision avoidance, are typically more urgent than the formation maintenance objective. As a result, it is necessary for the formation to compromise (i.e., be compliant) with these individual tasks, highlighting the need for swarm robots to behave compliantly. Inspired by the action principle of compliant control in physical robots, this paper proposes a distributed method that endows swarm robots with compliance. From the perspective of an individual robot, the method enables each robot to achieve its local tasks, allowing it to adapt to its environment. At the swarm level, the approach facilitates a compromise between formation maintenance and individual tasks, mitigating conflicts between individuality and collective objectives. Consequently, the swarm behaves compliantly, autonomously adjusting its formation shape. Finally, experimental results demonstrate the effectiveness of the proposed method, showing its ability to enhance the flexibility and adaptability of swarm formations.

Keywords: Multi-agent Systems
Abstract: Unmanned Aerial Vehicle (UAV) swarm systems are playing an irreplaceable role in various fields. However, the security challenge of UAV network has become increasingly prominent. In this paper, a lightweight secure access scheme based on physical unclonable function (PUF) for UAV swarm systems is proposed. Firstly, the UAV formation network is modelled and the security requirement is analyzed. Then, a lightweight authentication and key agreement protocol based on PUF is designed for UAV formation network to ensure access security. To verify the effectiveness of our scheme, the communication and computing costs of it are compared to other protocols.

Keywords: Optimal Control, Adaptive Control, Multi-agent Systems
Abstract: To address the nonlinear multi-agent formation tracking problem with input constraints and unknown communication faults, this paper develops a novel adaptive distributed observer-based model predictive control (MPC) method. The design integrates adaptive distributed observers into local control systems to estimate the leader’s state, dynamics, and desired displacement. By utilizing these observed information to construct local references, the original distributed formation tracking problem is decomposed into fully localized tracking control tasks, efficiently handled by local MPC controllers. This design enhances resilience against communication faults while simplifying the distributed MPC formulation.

Keywords: Optimal Control, Linear Systems, Motion Control
Abstract: This paper focuses on the design of a controller for vehicle with active suspension represented by a half car model, with aim to increase the crew comfort. The dynamics of the system is described by state-space equations, which are then used for the synthesis of the LQGI controller. Subsequently a PSO structure for fine tuning of the controller is created. The performance of the proposed controller is verified by MATLAB-Adams co-simulation. For the evaluation of the achieved results, the criteria for ride safety and comfort of car suspension according to ISO 8608 have been used. This work demonstrates a significant improvement in ride comfort and safety when using the proposed active suspension control.

Keywords: Optimal Control, Control Applications
Abstract: The fast dynamic characteristic of hydraulic servo systems poses a great challenge to the physical implementation of model predictive control (MPC). To address the online iterative solving of the optimal control problem, this paper proposes an efficient convex optimization pattern to obtain the optimal control variables in the shortest possible control period, thus supporting the system tracking performance and safety. An optimal control problem of the hydraulic servo system MPC is established based on the considered mechanism model. The convexity of the optimization model is proved, which makes the global convergence valid to avoid complex solving. As a model transformation prerequisite, the positive definiteness of the Hessian matrix is elucidated. Then, a conversion method from the MPC quadratic programming form to the non-negative least squares form is presented, which enables simpler implementation programming and more efficient optimization solving. Finally, some numerical experiments verify the effectiveness and merits of the proposed pattern.

Keywords: Control Applications, Process Automation, Optimal Control
Abstract: In the coal coking process, coal blending refers to the procedure of mixing different individual coal types from storage bins in specific proportions. The system controls the mass flow rate of each coal type to maintain the blending ratio, which is essential for enhancing production efficiency. Due to the uneven coal quality in storage bins, blockage at feeding ports and mechanism of coal flow monitoring, the coal blending system becomes a time-delayed and uncertain system. Existing control schemes fail to account for the impact of discharge port blockages on the maximum coal flow output. To address this, this paper proposes a two-layer data-driven model predictive control (MPC) framework. Model parameters are updated by using historical data and coal flow setpoints are recalibrated under constraints in the upper layer, while an MPC based on input-mapping method is solved in the lower layer to decrease the effects of the unknown but bounded uncertainties in system parameters. MATLAB simulations validate the effectiveness of the proposed method.

Keywords: Adaptive Control, Optimal Control, Factory Modeling and Automation
Abstract: The increasing demand for highly customized complex products, such as aircraft and aero engines, in dynamic global production environments has brought great challenges to manufacturing enterprises and supply networks. To tackle the problem of dynamics and networked control among distributed heterogeneous manufacturing resources, a hierarchical timed colored Petri net (HTCPN)-based self-adaptive optimal control (SOC) method is proposed for multi-level collaborative manufacturing networks. In contrast to existing HTCPN models, an industrial dataspace is designed to interoperate large-scale, multi-source, and heterogeneous real-time data, which provides manufacturing processes with data subspaces dynamically. To achieve SOC, the corresponding optimization problem is solved by a tailored multi-objective ant colony optimization (ACO) algorithm. A case study based on a Chinese aircraft manufacturer demonstrates the effectiveness and efficiency of the proposed method in reducing cost, time, and energy consumption. This paper potentially enables discrete manufacturing enterprises to implement SOC in multi-level manufacturing networks.