In this article, an adaptive distributed predefined-time and bounded (PTB) sliding mode controller based on a novel super-twisting-like estimator is proposed for the formation control problem of underactuated uncrewed surface vehicles (USVs) subject to unknown disturbances. To leverage the advantages of the super-twisting estimator and achieve ultimate PTB stability in the formation control system, a novel distributed predefined-time super-twisting-like estimator is first proposed. This estimator achieves real-time estimation of the leader's position and velocity within a predefined-time by collecting the neighboring USVs' estimates of the leader's position and velocity through switching interaction topologies, and it does not require the leader's acceleration information. Next, an improved nonsingular PTB sliding surface is designed. By integrating this sliding surface with adaptive control techniques, a distributed PTB sliding mode formation controller is designed for underactuated USVs subject to unknown disturbances. Finally, the PTB stability of the formation control system is rigorously proven. Numerical simulations of both fixed and time-varying formations validate the effectiveness and robustness of the proposed method.

Multi-UAV collaborative interception decision-making is a critical task in uncrewed aerial vehicle (UAV) applications. The core challenge lies in achieving real-time decision-making and precise coordination among multiple agents in highly dynamic and uncertain complex environments to successfully intercept highly maneuverable targets. However, existing methods still fall short in terms of rapid adaptability for interception decision-making, and the generalization performance of traditional reinforcement learning (RL) on new tasks needs improvement. To address these limitations, this article presents a novel meta-RL-based approach for multi-UAV interception decision-making. First, a task allocation method based on greedy strategy is proposed to achieve the rapid assignment of interception targets. Then, we design a task inference model that employs a probabilistic encoder to extract the shared structure across tasks from contextual information, thereby generating a latent variable to represent the task and enhance the training of the algorithm's network. Meanwhile, the proposed method is integrated with RL to improve its generalization capability in new tasks, enabling fast decision-making for the interception task. Finally, the comparative results against baseline algorithms validated the superiority of our method.

This article studies the trajectory tracking problem of unmanned surface vehicle(USV), where the USV is required to follow a predefined trajectory accurately. A hierarchical two-layer control architecture is proposed, consisting of an event-triggered model predictive control (ET-MPC) in the upper layer and a discrete PID controller in the lower layer. In particular, the ET-MPC reduces computational load and communication frequency by performing optimization only when the state error exceeds a predefined threshold or the cost function decrease condition is violated. The lower-layer PID loop applies feedforward compensation to decouple linear and angular velocities of the USV and uses a discrete-time difference model to account for the errors caused by actuator motor inertia and nonlinearity. In this way, the ET-MPC method can effectively balance the computational efficiency and performance of the controller. Finally, the hardware experiments are conducted to verify the effectiveness of the proposed method.

This paper addresses the challenges of model mismatch and hyperparameter sensitivity in path following control for multiple unmanned surface vehicles (USVs) operating in dynamic marine environments. We propose a synergistic framework integrating a Multi-Innovation Extended Kalman Filter (MI-EKF) and a Score-Driven Trust-Region Bayesian Optimization (Score-Driven TuRBO) to enhance the performance of Distributed Model Predictive Contouring Control (DMPCC). First, the MI-EKF algorithm dynamically estimates unmodeled dynamics by fusing multi-innovation observation residuals, enabling online compensation for the USV’s dynamic model used in DMPCC. Second, an improved Trust-Region Bayesian Optimization is developed with a dynamic scoring mechanism that adaptively allocates candidate points and adjusts scaling factors based on historical improvement rates, achieving efficient global optimization of DMPCC hyperparameters. Comparative simulations validate the efficacy of the proposed method.

This paper aims to propose a new event-triggered distributed model predictive control framework based on an adaptive dual-horizon (prediction and control horizon) mechanism for multi-agent cyber-physical systems (CPS) with additive disturbances. Firstly, the control horizon is introduced into the constrained optimal control problem (OCP) of multi-agent CPS to reduce the number of independent variables and improve the rapidity of the algorithm. Secondly, a periodic event-triggered mechanism with Zeno-free behavior is designed based on state error information, which can more efficiently reduce the frequency of solving OCP. Then, adaptive prediction horizon and control horizon contraction mechanisms are designed, effectively reducing the computational complexity of solving OCP for the controller at the triggering moment. In addition, sufficient conditions are provided through theoretical analysis to guarantee the recursive feasibility of algorithm and the stability of the closed-loop system. Finally, the effectiveness of the proposed algorithm is verified through simulation and experimentation example of networked multi-robot system, and the results showed that the algorithm can effectively reduce communication resource consumption and computational complexity in solving OCP while ensuring the expected cooperative control effect.

This paper studies the problem of multi-agent trajectory prediction in crowded unknown environments. A novel energy function optimization-based framework is proposed to generate prediction trajectories. Firstly, a new energy function is designed for easier optimization. Secondly, an online optimization pipeline for calculating parameters and agents’ velocities is developed. In this pipeline, we first design an efficient group division method based on Frechet distance to classify agents online. Then the strategy on decoupling the optimization of velocities and critical parameters in the energy function is developed, where the slap swarm algorithm and gradient descent algorithms are integrated to solve the optimization problems more efficiently. Thirdly, we propose a similarity-based resample evaluation algorithm to predict agents’ optimal goals, defined as the target-moving headings of agents, which effectively extracts hidden information in observed states and avoids learning agents’ destinations via the training dataset in advance. Experiments and comparison studies verify the advantages of the proposed method in terms of prediction accuracy and speed. Note to Practitioners—Autonomous robots and vehicles are rapidly integrated into social life and industry, and the scenarios that robots work with multiple people or other moving objects in a crowded environment such as streets and factories will be quite common. One of the most important problems for the robot to solve is the real-time and accurate trajectory prediction of multiple agents around itself to ensure safe navigation. However, existing methods either require prior information to train models or critical parameters in advance or have insufficient prediction accuracy, which are not suitable for robot safe navigation in real applications. In this paper, we investigate the real-time multi-agent trajectory prediction problem for a robot in crowded unknown environments. To obtain the accurate predicted trajectories in real-time, we propose a new energy function optimization-based framework to forecast multi-agent trajectories in crowded unknown scenarios. This framework utilizes the observed data to infer unknown information without the dataset and optimizes the trajectories very efficiently, which can be adopted for robot motion planning and navigation in real-world environments.

This article investigates the problem of real-time task assignment with heterogeneous agents while considering resource constraints. A hierarchical reinforcement learning-based(HRL) method to address the problem has been proposed. Unlike most existing studies, the method is to assign the tasks to agents such that the resource consumption is minimized while respecting the resource constraints and realizing task distribution balance among agents. At the high level, the real-time task assignment problem within a heterogeneous multi-agent system is formulated as a sequence optimization model which considers multiple constraints. At the low level, a novel reinforcement learning-based hierarchical framework that decomposes the large-scale task assignment problem into the target assignment layer and resource assignment layer is developed. In the first layer, a task assignment strategy closer to the optimal one in real-time by reducing the dimension of the DQN state-action space and learning from execution is produced. In the second layer, a novel resource assignment method is designed to minimize resource consumption and reserve as many agents as possible to handle new tasks by optimizing the resource distribution among agents. Simulation experiments in multi-UAV collaborative emergency material delivery demonstrate that the proposed method can generate high-quality solutions for various problem scales and greatly improve resource balance and real-time performance.

Mean field reinforcement learning (MFRL) addresses the problem of dimensional explosion for large-scale multiagent systems. However, MFRL averages the actions of neighbors equally while discarding the diversity and distinct features between individuals, which may lead to poor performance in many application scenarios. In this article, a new MFRL algorithm termed temporal weighted mean filed Q-learning (TWMFQ) is proposed. TWMFQ introduces a temporal compensated multihead attention structure to construct the weighted mean-field framework, which can sort out the complex relationships within the swarm into the interactions between specific agent and the weighted virtual mean agent. This approach allows the mean Q-function to represent the swarm behavior more informatively and comprehensively. In addition, an advanced sampling mechanism called mixed experience replay is established, which enriches the diversity of samples and prevents the algorithm from falling into local optimal solution. The comparison experiments on MAgent and multi-USV platform justify the superior performance of TWMFQ across different population sizes.

The complex and precise mechanical mechanism of quadruped robots is prone to faults, which brings challenges to the reliability and stability of the system. Therefore, it is of significant to develop the fault diagnosis method for quadruped robots, which can provide effective fault information for active fault-tolerant control. In this article, we propose a novel fault detection and isolation method for quadruped robots based on convolution neural networks, gated recurrent units, and attention networks, which can detect and isolate joint faults in real time. The proposed method can automatically learn meaningful high-level spatial and temporal features from sensors data. The effectiveness of the method is verified by the Laikago robot compound fault data.

In this article, we present a novel and robust registration method for autonomous underwater vehicles (AUV) using the forward-looking sonar (FLS). Due to the sparsity and repeatability of the underwater environment, and the loss of elevation angle, the FLS is difficult to consistently provide reliable information. The feature extraction and matching are the main difficulties encountered in registrations. This motivates us to propose a method to deal with various underwater environments and provide reliable information to improve the robustness and accuracy of the registration. First, the sonar images are preprocessed to highlight the region of interest and transformed. Next, a sonar image sequence is used for preliminary registration. Last, a combinatorial method including a feature-based region selection and a region-based registration is proposed. The validation is based on datasets and real-world experiments. Comparison studies verify the effectiveness of the proposed method.

This paper studies the problem of formation control for multiple quadrotors based on the model predictive contouring control (MPCC). A novel distributed multi-quadrotor formation control method is proposed which requires the quadrotors communicate the cooperative parameter and position to neighbors. In particular, the additional state in MPCC is selected as the cooperative parameter that is used to keep formation while following the reference path. In addition, an artificial potential field function is constructed to avoid collision. Finally, the simulation experiment is conducted to verify the effectiveness of the proposed method.

This article studies the distributed robust optimal control for discrete-time linear multiagent systems (MASs) with parametric uncertainties, where digraphs that only contain a directed spanning tree are allowed. Using the linear quadratic regulator approach, an optimal control protocol is presented. The presented controller is fully distributed, since the global information of graphs is unneeded for the design and implementation of the presented controller. The global performance index of MASs can be minimized by using the presented control protocol, and the optimal solution is independent with the information of parametric uncertainties. Finally, some simulated examples are provided to show the effectiveness of the proposed approaches.

This article is concerned with the event-based model-predictive control (MPC) problem of nonlinear cyber-physical systems with control constraints. A novel dual self-triggered MPC strategy is proposed to significantly reduce communication loads. In particular, two triggering mechanisms with two different optimal control problems are appropriately designed by considering whether the system state belongs to the terminal set. We prove that the strategy guarantees the algorithm feasibility and the closed-loop stability if the controller parameters satisfy the proposed conditions. Simulation and comparison studies verify effectiveness and advantages over the existing results.

The event-triggered control is a promising solution to cyber-physical systems, such as networked control systems, multiagent systems, and large-scale intelligent systems. In this paper, we propose an event-triggered model predictive control (MPC) scheme for constrained continuous-time nonlinear systems with bounded disturbances. First, a time-varying tightened state constraint is computed to achieve robust constraint satisfaction, and an event-triggered scheduling strategy is designed in the framework of dual-mode MPC. Second, the sufficient conditions for ensuring feasibility and closed-loop robust stability are developed, respectively. We show that robust stability can be ensured and communication load can be reduced with the proposed MPC algorithm. Finally, numerical simulations and comparison studies are performed to verify the theoretical results.

In this paper, an event-triggered distributed receding horizon control algorithm is proposed for dynamically coupled continuous-time linear systems. First, an event-triggered rule that is realized by monitoring the error between subsystem state and its optimal prediction is designed, and an extra constraint that restricts the discrepancy between each subsystem's assumed and predicted state and input trajectories is introduced to local optimization problems. Combined with the triggering rule and the extra constraint, the mutual disturbances caused by dynamical coupling are bounded and the inter-event time is lower bounded to avoid the Zeno behavior. Based on this, the algorithm feasibility and closed-loop stability are rigorously studied, and sufficient conditions for guaranteeing them are established. Finally, simulation studies are conducted to verify the theoretical results.

This paper investigates the issue on how to utilize neighbor information in the distributed receding horizon control (RHC)-based consensus problem for first-order multiagent systems. The distributed RHC-based consensus problem is first formulated in a general framework in terms of using neighbor information. Based on the framework, a sufficient condition on utilizing neighbor information to ensure consensus is developed for the finite horizon case. For the infinite horizon case, a necessary and sufficient condition is proposed, and the best way of using neighbor information to achieve fastest convergence rate is also presented. It is shown that: 1) the way of utilizing neighbor information plays an important role in reaching consensus; 2) the parameter that ensures consensus is related with the network topology; and 3) the best convergence rate in consensus can be attained if the neighbor information is appropriately utilized. Simulation studies verify the proposed theoretical results.