Tingting Yang

This paper investigates the predefined-time (PT) formation control for heterogeneous vehicle systems with general narrow tunnel constraints and non-affine actuator faults. Unlike conventional barrier Lyapunov function-based approaches that rely on specific constraint forms, a novel funnel-based control framework is developed to directly accommodate general curvilinear constraint boundaries. To address the explosion of complexity issue, an enhanced nonlinear filter is proposed, which reduces parameter dependence while improving filtering accuracy. By combining funnel control, coordinate transformations, and Butterworth low-pass filter, an adaptive broad fuzzy neural network control strategy is constructed to ensure PT stability of the closed-loop system. Besides, the proposed approach guarantees safe vehicle operation under narrow-tunnel constraints while keeping formation errors within prescribed performance funnels. Real-time experiments on a Speedgoat platform validate the effectiveness and feasibility of the proposed method.

Reliable detection of surface defects on metal materials is critical for ensuring product integrity and maintaining high manufacturing standards. However, existing object detection models often struggle with (a) subtle defect representation, (b) classification–localization misalignment, and (c) severe class imbalance, particularly in industrial scenarios. To address these challenges, we propose CUE-YOLO, a task-driven architectural refinement of the YOLOv11 framework, specifically tailored for efficient industrial surface defect inspection. Specifically, the novel C3K2 with Adaptive Sampling Attention (C3K2-ASA) module introduces an attention-guided feature refinement strategy to amplify subtle and fine-grained defect patterns. The designed Unified Task-Aligned Head (UTAH) mitigates task misalignment by harmonizing classification and localization branches through a unified prediction structure. Furthermore, the proposed Exponential Moving Average SlideLoss (EMA-SlideLoss) dynamically adjusts IoU thresholds to provide adaptive supervision, particularly for low-confidence detections and instances from underrepresented classes, thereby improving robustness in the presence of class imbalance. Experiments on the challenging GC10-DET dataset demonstrate that CUE-YOLO achieves a state-of-the-art performance of 74.2% mAP, outperforming the YOLOv11 baseline by 11.9%. Cross-dataset evaluations on aluminum and NEU-DET further verify the generalization ability of the proposed method across diverse industrial domains.

This article investigates the low-complexity predefined-time (PT) optimal formation control (FC) of multinonholonomic mobile robots (MNMRs) with collision avoidance safety requirements. First, by applying the funnel technique, the collision avoidance and communication range constraints between adjacent robots are uniformly transformed into formation error constraints, thereby eliminating the complex barrier Lyapunov functions in existing methods and reducing the complexity of controller design. Then, to mitigate the complexity explosion problem, an enhanced nonlinear filter with adaptive parameters is designed, which reduces the computational load while improving the convergence performance of the system. On this basis, the low-complexity safe FC scheme is constructed using broad fuzzy neural networks (BFNNs) and reinforcement learning techniques, achieving PT stability, optimality, avoiding collisions, and no overshoot behavior in formation errors. Finally, the effectiveness and feasibility of the proposed control method are verified on the Speedgoat real-time target machine.

Adaptive fuzzy predefined-time (PT) funnel formation control is investigated for uncertain third-order heterogeneous vehicle platoon systems (HVPSs) with collision avoidance, maintaining communication, asymptotic tracking, performance constraints (pre-specified transient behavior), actuator faults and external disturbance. Unlike the typical HVPSs results, the complete prior knowledge of the system parameters is not required. Firstly, to address the “explosion of complexity” challenge, an enhanced nonlinear PT filter is presented, which reduces the requirement of design parameter and improves the precision of linear filter. Then, an adaptive control strategy is proposed via backstepping technique and fuzzy approximation theory, which can realize the practical PT stability (PPTS) of HVPSs, and the asymptotic convergence of spacing errors is guaranteed by introducing smoothing function while meeting the constraint imposed by the performance funnel. Importantly, the complex Lyapunov function often associated with barrier Lyapunov function (BLF) technique can be avoided in this work. Finally, the validity of the theoretical results is verified by simulation of HVPSs under the predecessor-following communication topology.

Adaptive cooperative output regulation (COR) with funnel constraints is study for uncertain nonlinear multiagent systems (MASs) with nonsymmetric input saturation. First, an improved distributed predefined-time-based observer is presented to estimate the exogenous signal and the proposed controller based on the observer realizes the predefined-time (PT) convergence of the estimate errors. To tackle the “explosion of complexity” issue, a modified nonlinear filter is presented. Then, an adaptive algorithm utilizing the dynamic surface technique is developed to ensure practical PT stability of the controlled system (the settling-time can be preset arbitrarily), while meeting the funnel constraints. Notably, the inverse transformation function of the transformation error is not involved in the algorithm, which makes up for the deficiency of prescribed performance control. Meanwhile, the algorithm removes the assumption that the system functions are known, which reduces the conservatism of the algorithm. Finally, the validity of the theoretical findings is verified by simulation examples.

Adaptive consensus fixed-time (FT)-based prescribed performance is studied for nonlinear high-order multiagent systems (MASs) with nonstrict-feedback form and event-triggered communication. To address the “explosion of complexity” issue, an enhanced nonlinear FT filter with tanh and cubic functions is proposed. Under the backstepping and prescribed performance control (PPC) framework, a distributed event-trigger condition and an FT dynamic surface control (DSC) scheme are developed under directed topology. The main advantage of the developed algorithm is that the continuous communication between adjacent agents is not required, which can reduce the data communication and the controller update frequency. Meanwhile, the Zeno behavior can be effectively avoided. Simulation examples show that the developed method can significantly reduce the control effort on the premise of ensuring the practical FT stability (PFTS) of the controlled system and the tracking errors meeting the prescribed performance constraints.

Adaptive containment control with funnel performance and practical predefined-time (PT) convergence is investigated for uncertain nonlinear heterogeneous multiagent systems (MASs) with multiple leaders. Unlike the existing funnel literatures, the considered system has unknown sensor and actuator failures simultaneously, and no assumptions are imposed on the system function. In particular, different adaptive laws are introduced to reduce the effects of sensor and actuator faults. An improved nonlinear filter is employed to avoid the “explosion of complexity” issue and realizes the convergence of filter errors within the PT. Additionally, the control singular issue is avoided by using L’Hospital’s rule. Subsequently, it is shown that the proposed controller can realize that all the signals are practical PT bounded, the followers converge to the convex formed by multiple leaders and the containment errors remain in the prescribed funnel region. Finally, simulation examples are performed to verify the effectiveness of the proposed algorithm.

This article endeavors to discuss the adaptive fuzzy funnel consensus tracking issue for uncertain nonlinear multi-agent systems (MASs) with unknown dynamics, non-affine nonlinear faults (NANFs), high powers (possibly different), performance constraints and fixed-time convergence. The performance constraints include pre-specified transient and steady-state tracking error behaviors. Firstly, to overcome the “explosion of complexity” issue, an enhanced nonlinear filter with adaptive parameter estimator and cubic function is constructed, which can ensure that the boundary layer errors fixed-time convergence and fully compensate their effects. Then, under the backstepping and separation function framework, a distributed algorithm is presented via Butterworth low-pass filter (BLPF) technique, funnel method and dynamic surface control (DSC) technique. The results confirm that the proposed algorithm can not only guarantee the considered MASs is practical fixed-time stability (PFTS), but also ensure the asymptotic convergence of the consensus errors and meet the constraints imposed by the funnel. Moreover, compared to the barrier Lyapunov function (BLF) technique for MASs, the considered system is more versatile, and the complex control algorithm can be avoided. Finally, simulation results are performed to verify the effectiveness of the developed theoretical findings. Note to Practitioners —It is well known that some underactuated and weakly coupled mechanical systems cannot be accurately represented by traditional low-power models. Therefore, it is of great practical significance to study high-power systems to better understand and control this kind of complex systems. Besides, in actual complex systems, actuator failures are inevitable, for example, in the process of signal processing, image processing, communication and wireless transmission, etc., may suffer from non-affine faults. Furthermore, to optimize system performance, some engineering applications often require to constraint the convergence time, the maximum error overshoot and the minimum steady-state error. Thus, the distributed fixed-time funnel performance constraint control is studied for MASs with NANFs.

Adaptive fuzzy practical predefined-time leader-following consensus with funnel performance is investigated for nonlinear multi-agent systems (MASs) subject to fuzzy dead zone (FDZ). The dead zone model is uncertain and fuzzy, which makes the developed scheme different from the existing work of funnel performance. Firstly, to avoid the “explosion of complexity” problem, an improved nonlinear predefined-time filter is introduced and the convergence time of filtering errors can be preset. Subsequently, a distributed controller is constructed via the dynamic surface technique, which can not only reduce the influence of FDZ nonlinearity on the system performance, but also realize the considered system is practical predefined-time stability (PPTS), while satisfying the transient and steady-state performance imposed by the funnel. Furthermore, unlike the literature for solving funnel-like performance, the inverse transform function and the complex Lyapunov function are not involved in the proposed method, which can avoid the potential singularity issue and complex control law. Finally, simulation examples are performed to verify the effectiveness of the developed algorithm. Note to Practitioners —Complex engineering systems are difficult to control with a single component, so the cooperative control of multi-agent systems (MASs) has attracted more and more attention due to its wide range of applications, such as multi-motor synchronization, spacecraft formation flying, etc. Consensus, as one of the important issues in MASs cooperative control, has become a research hotspot. Dead-zone (DZ) nonlinearity is one of the most common constraints in physical systems, which often occurs in electric servo motors, hydraulic servo valves, etc. In the existing literature, the DZ feature is definite and accurate. However, due to the physical characteristics of the actuator and the influence of the environment, there are often uncertainties in the DZ parameters. On the other hand, performance behaviors (convergence rate, maximum overshoot and maximum steady state error, etc.) often required to be pre-specified in practical applications. Therefore, the funnel-based leader-following consensus control is studied for MASs with fuzzy dead zone (FDZ) and predefined-time convergence.