Hybrid brain-computer interfaces combining electroencephalography and functional near-infrared spectroscopy utilize neurovascular coupling to improve decoding. Current fusion strategies rely on shallow aggregation or complex attention, which struggle to capture complex temporal dynamics and complex dependencies in long sequences. Moreover, static models lack online calibration for sensor failures or non-stationary noise, causing significant performance drops. We propose the Multi-Scale Bidirectional Mamba with Test-Time Adaptation (MS-BiM-TTA) framework. The architecture employs a multi-scale frontend for heterogeneous feature extraction and bidirectional state-space models for efficient long-range intra-modal and inter-modal modeling with linear complexity. To ensure robustness, a customized test-time adaptation workflow utilizes two-level entropy filtering for update safety and mutual information sharing for autonomous representation reconstruction. Results show that MS-BiM-TTA significantly enhances accuracy and demonstrates superior resilience to modality missingness in both binary and fine-grained tasks.
Knowledge diversity and catastrophic forgetting are two significant challenges in sustaining long-term collaborative learning within heterogeneous vehicle networks. These challenges become more notable when vehicles have varying sensing capabilities, computational resources, and domain expertise, which can result in fragmented learning and unstable knowledge retention. To tackle these issues, we propose a dual-level game-theoretic framework for UAV-assisted heterogeneous vehicle networks. We introduce a new metric called Utility-of-Information (UoI) that quantifies knowledge diversity, retention, and consolidation in a comprehensive manner. The framework comprises a lower-level coalition formation game, where vehicles self-organize into “teacher-student” coalitions based on their UoI profiles, and an upper-level UAV resource allocation game, where vehicle coalitions compete for limited communication resources. To solve the associated resource optimization problem under dynamic conditions, we develop a unified reinforcement learning framework utilizing Proximal Policy Optimization (PPO). Extensive simulations and testbed validations demonstrate that the proposed approach effectively manages knowledge diversity, mitigates catastrophic forgetting, and supports long-term collaborative learning in real-world UAV-assisted vehicle networks.
Large Language Models (LLMs) are becoming key enablers in adaptive and autonomous systems, particularly under the paradigm of Industry 5.0, where human-centric design and generative Artificial Intelligence (AI) technologies are increasingly deployed. However, the widespread of LLMs raises serious intellectual property concerns, especially in few-shot learning scenarios where model customization is achieved through continuous prompt tuning. Traditional watermarking methods fail to protect such models due to their limited data access and fixed model parameters. To address this challenge, we propose a novel backdoor-based watermarking framework tailored for continuous prompt learning in few-shot settings. Our method leverages semantic-aware trigger generation and adaptive trigger assignment to embed robust and invisible behavioral watermarks without compromising model performance. Specifically, we introduce a semantic alignment mechanism to generate and filter watermark triggers. We also present an adaptive strategy to assign optimal triggers based on decision boundary proximity. Experimental results across various NLP tasks demonstrate high watermark detection accuracy, minimal impact on model utility, and resilience to adversarial removal. This work contributes a practical and effective approach to copyright protection for generative AI models in Industry 5.0 systems.
Deep Neural Networks (DNNs) are widely used in intelligent applications, driving increasing computational demands on GPUs. However, modern GPU multitasking scheduling algorithms fail to effectively balance real-time task performance and resource utilization, especially under dynamic workloads with highly variable DNN computational demands. The complex and workload-dependent execution times of DNN kernels often lead to inefficient resource allocation, degraded system throughput, and missed real-time constraints. To address these challenges, we propose Flexible Parallel Orchestrator (FlePo), a GPU multitasking scheduling framework designed to optimize resource utilization and maintain real-time task performance within acceptable limits for soft real-time systems. FlePo integrates two key techniques: Adaptive Padding Dispatch (APD), which dynamically schedules best-effort tasks while leveraging the predictable execution characteristics of DNN kernels to maintain real-time predictability; and Dynamic Parallel Fusion (DPF), which employs kernel fusion to create computational isolation, reducing interference in parallel job execution. By combining offline profiling with online adaptation, FlePo efficiently responds to workload variations. We evaluate FlePo on two heterogeneous GPU platforms, NVIDIA Tesla V100 and AMD MI50, achieving up to a 50% increase in throughput while keeping real-time overhead below 2%. Our work enhances GPU multitasking in dynamic environments, with potential applications in autonomous driving, smart homes, and intelligent healthcare.
Human-object interaction detection methods usually adopt a multi-task framework, including a feature learning backbone and two sub-tasks: instance detection and interaction classification. These two sub-tasks shared the same image representation. However, different sub-task requires different image information. For instance detection, detector focused on the feature of local region while interaction predictor required the feature on bigger receptive field. To solve this problem, a new HOI detection framework is designed to select the appropriate representation for each sub-task. Specially, the local feature is learned to predict the center points of instances. For the interaction detection, Transformer is chosen to extract the context information to improve the accuracy. In the matching stage of human-object pair, the offsets from human and object to the interaction point are predicted to obtain more accurate interaction pair. Finally, the experimental results show that compared with the existing algorithms, the proposed method achieves the better performance.
This is a corrigendum for the article “LyDRL: Lyapunov-guided Deep Reinforcement Learning for Stable Task Offloading in Connected Autonomous Vehicles” published in ACM Trans. Autonom. Adapt. Syst. 20, 3, Article 24 (September 2025), 29 pages.
In autonomous and adaptive Industry 5.0 systems, sensors on various devices continuously generate high-throughput time series data that are characterized by intensive concurrent requests, high-frequency sampling, and multisource heterogeneity. These complex and large-scale data streams are prone to containing various anomalies caused by noise interference, equipment faults, or operational issues. Therefore, robust and accurate anomaly detection techniques are essential for ensuring the stability and safety of autonomous and adaptive systems. Multivariate time series anomaly detection tasks in autonomous Industry 5.0 systems face two major challenges: insufficient representations of spatio-temporal dependencies and the inadequate detection of diverse anomaly patterns. To address these challenges, we propose an unsupervised anomaly detection framework for multivariate time series that involves Spatio-Temporal Learning and Diffusion Generation (STDiff). First, the Denoising Diffusion Probabilistic Model (DDPM) is adopted as the underlying architecture. The local features of the given time series are learned by multiscale spatio-temporal convolution, while a linear attention mechanism is incorporated to analyze long-term time series. Second, given the long-term context-dependent nature of multivariate time series, we propose a Temporal Feature Memory (TFM) that integrates the memory features encoded at multiple scales with the decoded features. It enhances the ability of the model to preserve long-term temporal patterns. Third, we design a Spatial Interaction Awareness (SIA) for multisource and heterogeneous industrial time series in autonomous systems. It adaptively learns the inter-variable correlations within multivariate time series through Graph Attention Networks (GATs), and the spatial feature representation is enhanced by residual connections. Finally, experimental results obtained on multiple real-world industrial datasets show that the proposed method outperforms the baseline methods and exhibits excellent anomaly detection performance.
Federated learning is an emerging paradigm that addresses privacy concerns by training on decentralized data, but it faces significant performance challenges due to factors like non-independent and identically distributed (non-IID) data and dynamic data distributions. These challenges are further amplified in federated class-incremental learning under realistic, non-monotonic class streams, where previously seen classes can intermittently reappear alongside new classes—making generative replay highly susceptible to training instability, mode collapse, and feature drift. To address these issues, we propose RAVEL-FCL , a generative replay-based framework for federated continual learning. Our approach integrates an improved generative model based on Rebooting ACGAN with multi-level feature alignment to ensure consistency, and employs Elastic Variational Continual Learning on the server to probabilistically regularize the global model and preserve past knowledge. Experimental evaluations show that our method achieves stable convergence and significantly outperforms state-of-the-art methods on standard benchmarks—for example, it improves accuracy by 5.27% over FedCIL on EMNIST-Balanced and by 1.60% and 1.27% over the strongest baselines on TinyImageNet and ImageNet-LT, respectively—while maintaining a forgetting rate below 5% across tasks. By jointly addressing feature drift and catastrophic forgetting, RAVEL-FCL provides a robust and privacy-preserving solution suitable for the adaptive and autonomous systems envisioned for Industry 5.0.
Safe exploration in reinforcement learning remains a critical challenge for safety-critical autonomous systems, where the typical trial-and-error learning process can lead to hazardous outcomes. While several existing approaches incorporate kinematic models or external knowledge to limit the exploration of unsafe behaviors, their effectiveness is significantly weakened in the presence of incomplete or sparse knowledge. This paper introduces a counterexample-guided reinforcement learning method that navigates safe exploration in autonomous systems without prior knowledge, even when safety and optimality conflict. Our method geometrically abstracts discrete and continuous state-space systems into compact, PAC-learnable models that capture safety-relevant information. We then generate probabilistic counterexamples of the safety requirement to regulate online exploration toward minimizing safety violations, relying on minimal offline counterexample-guided simulations. We further propose a novel belief-based regularization method to address the distributional shift between online and offline learning and to balance optimization and safety, ensuring conservative behavior with theoretical guarantees. Our evaluations demonstrate the effectiveness of the method in significantly reducing safety violations without compromising cumulative rewards when benchmarked against other Q-learning or actor-critic methods with unconstrained or safety-constrained exploration.
Designing effective human-robot interaction (HRI) for multi-Autonomous Guided Vehicle (AGV) systems in manufacturing remains a significant challenge. While existing modeling tools offer formal representations for system behavior or task flow, they lack support for explicitly modeling multimodal, bidirectional communication between humans and distributed autonomous agents. In this paper, we introduce HASIGN: Human-Autonomous System Interaction Graphical Notation in order to make the design space of human-multi-AGV interaction explicit and tractable. Developed through a Research through Design approach, HASIGN integrates agent roles, interaction modalities, and temporal intent communication into a unified and practical representation. We apply HASIGN to five diverse industrial case studies drawn from ongoing research and development projects. These cases demonstrate the notation's flexibility and domain suitability, while uncovering unexplored areas in the interaction design space. Rather than aiming to replace existing modeling tools, HASIGN complements them by focusing on human-centered communication in autonomous systems. This paper contributes a visual design tool for practitioners and researchers, and lays the foundation for further evaluation, standardization, and adoption in the context of human-centered autonomous intelligent systems (HCAIS).
WiFi-based human activity recognition (HAR) is a promising solution for enabling intelligent sensing in autonomic systems such as smart homes and industrial environments. Recent research has explored deep learning models for this task, yet two core challenges persist, i.e., signal degradation caused by environmental noise, and the difficulty of extracting discriminative features from heterogeneous time-frequency domains. To address these limitations, we propose a generative artificial intelligence (GenAI) framework that integrates a conditional diffusion model with a pixel-aware attention mechanism. The diffusion model enhances data quality by reconstructing clean channel state information (CSI) signals through a forward noise injection and a reverse denoising process. The pixel-aware attention module adaptively fuses multi-resolution features from short-time Fourier transform (STFT) and discrete wavelet transform (DWT) spectrograms at channel, spatial, and pixel levels, improving the representation of fine-grained activity patterns. To the best of our knowledge, this is the first work to apply a generative denoising diffusion with fine-grained pixel-level fusion for this task. We evaluate our model on four public datasets, i.e., SignFi, Widar3.0, UT-HAR, and NTU-HAR. Experimental results show that our model consistently outperforms existing methods, demonstrating strong robustness and generalization in both gesture and action recognition tasks.
Industry 5.0 represents a transformative paradigm that emphasizes synergy between human expertise, intelligent systems, and hyper connected cyber-physical environments. While this evolution fosters personalized automation and resilient production, it also amplifies the cybersecurity risks inherent in Industrial Internet of Things (IIoT) infrastructures. In this research, we present Aegis-5 a novel adaptive hybrid ensemble framework explicitly designed for intrusion detection in Industry 5.0-enabled smart manufacturing ecosystems. The proposed model integrates five diverse classifiers Random Forest, Gradient Boosting, XGBoost, SVM, and K-Nearest Neighbors using a dynamic weighting strategy guided by per-class precision, recall, and F1-score performance in real time. A meta-learner further synthesizes these predictions to enhance robustness against sophisticated and zero-day attacks. To ensure relevance and reliability, we evaluate the model using two benchmark IIoT datasets: IoT-23 and CIC-IoT 2023, both of which capture a broad spectrum of real-world industrial threats. Experimental results demonstrate that our framework achieves superior performance, with accuracy rates of 99.98% on IoT-23 and 99.95% on CIC-IoT 2023, coupled with precision (99.97%, 99.93%), recall (99.96%, 99.92%), and F1-score (99.96%, 99.93%) respectively., significantly reduces false positives, and adapts effectively to evolving attack behaviors. By aligning intelligent anomaly detection with the responsiveness and adaptability required by Industry 5.0, Aegis-5 offers a scalable, real-time, and practical cybersecurity solution for next-generation industrial systems.
Industry 5.0 is emerging as the next phase of industrial evolution, emphasizing human-centric manufacturing through close human–robot collaboration and the deployment of intelligent autonomous systems. As a representative example of such autonomy, person-following robots are typically implemented on differential-drive or omnidirectional mobile bases. However, certain tasks require Ackermann-steered robots, which face unique challenges due to limited maneuverability and the complexity of urban environments, often leading to target loss or navigation into non-drivable areas. To address these issues, we propose Auto-Follower, a person-following framework with enhanced perception and navigation capabilities. Auto-Follower integrates a vision–LiDAR servo tracker that fuses camera images with LiDAR points from a motorized rotating sensor, enabling 360° target perception. Instead of relying on a global map, the system employs real-time LiDAR-based local mapping for efficient path planning. In addition, an Iterative Radius Points Search (IRPS) method is developed to identify obstacle-free navigation goals when the target enters non-drivable regions, ensuring safe and continuous following. The framework has been validated extensively in both laboratory and urban environments and demonstrates robust, reliable performance, with strong potential for adaptation to diverse real-world person-following applications.
The successful deployment of Digital Twins for real-time optimisation of physical systems relies critically on highly accurate and efficient deep learning surrogate models. Ensuring these models meet performance and (HPO). While both have been traditionally posed as a pure-exploration bandit problem, we show that it fails to capture the unique, deterministic characteristics of Scientific Machine Learning (SciML) models underpinning ONets). We propose a non-stochastic multi-armed bandit with balanced exploration-exploitation as the proper setting for NAS in SciML and introduce BanditNAS, a novel algorithm that addresses three critical challenges absent from current approaches: (i) late convergence in high-capacity models exhibiting spectral bias, (ii) validation loss plateaus requiring optimiser switching and (iii) the deterministic (non-stochastic) nature of physics-based training data. We analyse BanditNAS's theoretical properties, proving improved regret bounds compared to adaptive adversaries and compare it empirically with state-of-the-art approaches across three representative SciML scenarios. Our results demonstrate setting-dependent performance: BanditNAS achieves up to 95% higher optimal selection rates when multi-stage fine-tuning is required (DeepONets with L-BFGS switching), approximately 50% improvement in late-convergence regimes (high-capacity PINNs) and comparable performance to HyperBand in moderately noisy environments, although underperforming in very large search spaces with high noise ( = 200 graph networks). Statistical significance testing confirms BanditNAS's superiority in two of three settings (p < 0.001), with competitive performance in the third when restricted to K <= 100. These findings establish BanditNAS as a viable and theoretically grounded approach for optimising SciML models where real-time accuracy, multi-stage training and computational resource constraints are paramount, while highlighting the importance of algorithm selection based on problem characteristics.
Autonomous systems in VEC-enabled ICVs face many challenges, such as self-organization, privacy breach risks, vehicle selection, and resource allocation. As a distributed training framework, Federated Meta-Learning (FML) provides a powerful tool for adaptive and efficient processing of vehicular tasks while securing vehicle data privacy in VEC-enabled ICVs. However, the high-speed mobility of vehicles leads to higher latency and communication interruptions. This article investigates the vehicle selection and resource allocation scheme, subject to the constraints on the number and the residence time of vehicles, the maximum transmission energy consumption, and the ratio of bandwidth resource allocation. It is proved to be a challenging mixed-integer nonlinear programming problem, and we formulate it as a Markov decision process (MDP). We proposed an adaptive Sum Tree-Deep Recurrent Q-network algorithm (ST-DRQN) to solve the optimal resource allocation. ST-DRQN employs an enhanced empirical selection rule and a proportional priority sampling method to address the problems of inefficient model training and slow convergence. Finally, we conducted experiments using intelligent cars equipped with Raspberry Pi to show the effectiveness of the proposed methodology. Experimental results demonstrate that ST-DRQN achieves adaptability and credibility among ICVs while reducing latency and energy costs incurred by long-term training of FML.
Mobile multimedia applications such as real-time video processing, augmented reality, and mobile gaming have raised high requirements for low latency and high efficiency. Edge-based autonomous systems have become a key technology for processing these application tasks. This article focuses on joint resource allocation and task slicing for mobile multimedia computing in edge-based autonomous systems. We propose an efficient resource allocation and task slicing strategy, aiming at the optimization of the overall utility of both edge servers and mobile devices simultaneously. We transform the resource allocation problem into resource pricing and purchasing behaviors. We present a Stackelberg game model and prove theorems for the existence of equilibrium and optimality. Based on the theorems, we design an algorithm namely G-RPTSS for resource purchasing and computation task slicing. Then, we employ Deep Reinforcement Learning (DRL) techniques in resource pricing and propose the DRL-ESRP algorithm which is capable of adaptively responding to dynamic computational scenarios in edge-based autonomous systems. Our scheme leverages the DRL technique for autonomous learning and policy adjustment. Simulation experiments, based on real-world scenario data, demonstrate the superior of our approach in learning efficiency and performance advantages to existing both non-DRL and other DRL algorithms.
