Role-based collaboration has proven to be a promising approach to addressing collaboration issues. Environments-classes, agents, roles, groups, and objects (E-CARGO) is a formalized model of it. Group multi-role assignment is core to such collaboration, yet it is seriously constrained by cooperation and conflict relationships between agents and roles—a key practical challenge existing research fails to clearly address. Traditional studies mostly adopt predefined matrices to depict agent-role interactions, while lacking empirical support. To fill this gap, this paper adopts three-way decision theory, introduces agent preference, and employs the Spearman rank correlation coefficient to measure agent-role relevance. With set thresholds, it identifies internal cooperation and conflict relations, and further applies the method to optimize group multi-role assignment for conflict elimination and cooperation promotion. This paper formulates necessary constraint conditions and designs corresponding verification algorithms. Experimental and comparative results verify the effectiveness and practicality of the proposed model.
The distributed event-triggered consensus problem of linear systems is studied in this article. Unlike the existing event-triggered adaptive consensus protocols, where the control gain is composed of an adaptive weight and a feedback matrix, this article proposes an integrated adaptive event-triggered control approach. The control gain matrix can be adaptively and intermittently updated according to the consensus behavior. It is proven that the developed event-triggered mechanism can reach consensus and avoid the Zeno behavior. Some simulation results are provided to illustrate the effectiveness of the proposed protocols.
Offshore wind power serves as a pivotal pillar of sustainable energy, with its operational and maintenance efficiency heavily dependent on effective human resource coordination. This paper proposes a dynamic personnel allocation optimization method based on hypergraph structures, constructing an RGF propagation model to characterize the many-tomany complex interactions between teams and tasks across multiple task categories. Unlike traditional graph models that can only represent pairwise relationships, the hypergraph framework captures group-level collaborations and spatiotemporal propagation effects, thereby more accurately simulating the dynamic evolution of personnel states. Furthermore, a reinforcement learning control framework based on Proximal Policy Optimization (PPO) is designed to achieve adaptive strategy optimization in uncertain environments. Numerical simulations and multibaseline comparative experiments validate the effectiveness of the proposed method: the PPO controller significantly reduces task completion time, personnel requirements, and overall costs, while demonstrating superior performance over traditional methods and other reinforcement learning algorithms in dynamic task environments. This framework provides a novel theoretical foundation for human resource scheduling and control optimization in large-scale offshore wind projects, offering substantial practical potential.
While reinforcement learning (RL) has advanced optimal control under limited resources, existing approaches often neglect two critical challenges in real-world engineering: switching communication topologies and time delays. This paper distinguishes itself by addressing the online adaptive optimal consensus control problem for multi-agent systems (MASs) that explicitly include these factors. The switching network topology is characterized by a Markov chain, capturing random changes, and a periodic switching signal, modeling scheduled variations. We first formulate a quadratic cost function to evaluate system performance and derive the associated Hamilton-Jacobi-Bellman (HJB) equation. A policy iteration (PI) method is established to solve this HJB equation offline. Leveraging this, a novel online adaptive optimal controller is proposed using a critic-actor neural network structure. This structure learns the optimal control policy in real-time without requiring knowledge of the system dynamics. The convergence of the policy iteration algorithm and the stability of the closed-loop system are rigorously analyzed. Finally, simulation examples are presented to demonstrate the efficacy and superiority of the proposed control mechanism.
People live in a society and cannot live alone. Social intelligence (SoI) is required for every human being, even though current Artificial Intelligence (AI) ignores it. Similar to the idea of SoI, systems, collaborative, collective, and team intelligence are also discussed and investigated. However, because the investigation is mainly from the viewpoint of humanity, it is hard to formalize a well-accepted concept. Thanks to the Environments – Classes, Agents, Roles, Groups, and Objects (E-CARGO) model and the methodology of Role-Based Collaboration (RBC), it is possible to specify a unified concept about such intelligences, called team intelligence. This article presents a real-world scenario to induce the requirement to state the intelligence of a team and ensures that team intelligence is supported by E-CARGO/RBC. Then, this paper proposes an indicator for team intelligence. By reviewing related work, it clarifies the terms of social, systems, collaborative, collective, and team intelligence, consolidating the contributions of the proposal.
With the increasing complexity of unmanned aerial vehicle (UAV) swarm mission scenarios, traditional physical testing methods face challenges in terms of cost, safety, and scalability, making efficient, reliable, and cost-effective verification methods essential. A hybrid virtual-real simulation system is presented, integrating physical UAVs, high-fidelity virtual UAV models, and a virtual confrontation environment capable of simulating complex scenarios that are difficult to deploy in reality. A gateway-based hierarchical communication architecture is designed for multi-platform data fusion, with time synchronization and latency compensation mechanisms introduced to ensure data consistency across virtual and physical platforms. Cooperative confrontation experiments are repeatedly conducted under node failures and varying swarm scales, demonstrating that the proposed system provides a feasible and cost-effective method for validating swarm cooperative algorithms in complex confrontation environments.
Autonomous air-combat decision-making for unmanned aerial vehicles (UAVs) faces a critical challenge: pure reinforcement learning (RL) suffers from severe cold-start problems in high-dimensional adversarial environments, while imitation learning (IL) alone cannot exceed human performance. This paper proposes a two-stage human-guided learning framework that integrates generative adversarial imitation learning (GAIL) for policy initialization and proximal policy optimization (PPO) for subsequent refinement. In Stage 1, GAIL bootstraps a competent policy from only 2.7 hours of non-expert demonstrations, after which the UAV masters the complete tactical sequence of pursuit, lock-on, and engagement with substantially fewer samples than pure RL baselines. In Stage 2, PPO elevates performance beyond human-demonstrator levels. Validation in the Harfang 3D Dogfight high-fidelity simulator shows a 2.3- fold improvement in kill success rate and a 30 percent reduction in time-to-kill compared to state-of-the-art RL methods. These results indicate that modest human guidance can effectively overcome the cold-start barrier in adversarial continuous control, offering a practical pathway toward robust autonomous combat systems. The central finding is that GAIL enables the neural network to acquire the correct engagement sequence— establishing radar lock prior to weapon release—a fundamental combat discipline that pure RL did not discover within the tested training budgets.
Access to pediatric dental care is often restricted by a lack of specialized dentists and the high cost of diagnostics, leading to delayed treatment. This study introduces a deep learning methodology for identifying pediatric or deciduous dental lesions in clinical images, offering two significant advancements. First, we developed a pediatric dental lesion dataset comprising 390 high-resolution images annotated with 2,114 detectable dental pathologies, ranging from small to moderate to severe lesions/tooth decay, which provides comprehensive resources for studying pediatric dental lesions. Second, we propose a novel deep learning framework that combines the Swin-Transformer with Mask region-based convolutional neural networks (R-CNNs) for precise lesion detection. This model is adept at identifying signs of various pediatric dental lesions, including patients with multiple disease sites. The Swin-Transformer component extracts features from image patches, capturing both local details and broader contextual information. Mask R-CNN then uses these extracted features to accurately identify and differentiate between cooccurring lesions within the same image. This approach addresses the challenge of diagnosing multiple lesion types that are present simultaneously, a common scenario in pediatric dentistry. To our knowledge, these contributions represent a novel approach for the detection of pediatric dental lesions. The testing of our model on the proposed dataset shows substantial improvements in both the accuracy and efficiency of image-based diagnoses, validating the effectiveness of our proposed method.
Data privacy and cybersecurity are more critical than ever in this digital era. As we rely more on technology to store and transmit personal and sensitive information, we must take precautions against cyber-attacks and breaches. Sensitive data leaks happen frequently and have increasingly worse consequences, it can result in significant damage or even privacy disclosure. This paper explores the intersection of these important areas, concentrating on creating and assessing an original solution known as the CyberGuard Suite. The main goal is to evaluate data privacy effects on cybersecurity by using an advanced browser security extension such as CyberGuard suite as a lens. CyberGuard includes features including HTML led content creation, cascading style sheets comprising friendly user interface, threat Intelligence for ad blocking and Redirection, plus Phishing detection attacks. CyberGuard suite has been compared with existing state of art and is found competitive enough in terms of key performance indicators. On average CyberGuard suite achieved 47 % faster page load time, 28% more resistant to Ad blocking and gained 17% improvement in terms of phishing detection.
One of the enduring challenges in education is how to empower students to take ownership of their learning by setting meaningful goals, tracking their progress, and adapting their strategies when faced with setbacks. Research has shown that this form of leaner-centered learning is best cultivated through structured, supportive environments that promote guided practice, scaffolded inquiry, and collaborative dialogue.
Appropriate bed temperature is essential for ensuring the stability of CFB operation. However, due to the complex physicochemical reactions involved in the combustion process and other factors, the bed temperature is influenced by a wide range of parameters, making it highly challenging to develop a model that can accurately predict the bed temperature. In this study, a physical information-dominated residual network model for bed temperature prediction in CFB is proposed. The method innovatively combines the mechanistic model with deep learning techniques. It integrates the physical model into the training process in the form of a loss function, which ensures the physical consistency of the model’s output and enhances the model’s interpretability. Validated with data from actual operating units, the proposed method demonstrates superior accuracy and reliability. Specifically, through numerous experiments, the model achieves the root mean square error (RMSE) of 5.833, mean absolute error(MAE) of 4.827, and mean absolute percentage error (MAPE) of 0.570%, showcasing industry-leading performance in prediction accuracy compared to other models.
This paper studies the global optimal consensus problem of unknown leader-follower multi-agent systems. The problem presents two primary challenges: the coupling between communication and control, and the expansion in computational complexity as the number of agents increases. To tackle these difficulties, a two-fold approach is proposed. First, computational decomposition is achieved by designing appropriate weighting matrices, and a reinforcement learning(RL)-based algorithm is developed to facilitate the learning of local optimal gains. Additionally, an information fusion method is introduced, enabling each agent to derive the optimal controller in a distributed manner. Moreover, a numerical simulation example is presented to illustrate the effectiveness of the proposed method.
Role-Based Collaboration (RBC) theory and its ECARGO model, composed of Environment, Classes, Agents, Roles, Groups, and Objects, have significantly transformed collaborative work and addressed a range of practical challenges. Existing literature has primarily focused on single-level role assignment, while recent studies have introduced Group Bi- Level Role Assignment (GBLRA), which tackles collaborative assignment problems in which the assignment of subordinate roles triggers the need for leader roles. However, GBLRA still suffers from several limitations, which can be summarized as follows: (1) it does not account for the fact that a role generation plan may produce multiple distinct leader roles at once; (2) it fails to set qualification boundaries for agents acting as leader roles or consider the impact of their competence on group performance; and (3) it overlooks the influence of ungrouped agents on group performance. To address these issues, this paper extends the GBLRA framework and proposes two enhanced models: Developed GBLRA-1 (DGBLRA-1) and DGBLRA-2. Experimental results demonstrate the effectiveness of the proposed models.
According to Kings Research [1] the global Wearable Market is projected to reach US$ 192.14 billion by 2030, growing at a CAGR of 15.8% from 2023 to 2030, with an increasing growth caused by the COVID-19. The industry and public sector are then pushing for innovative WCS solutions with high levels of dependability and trustworthiness that can efficiently operate in increasingly complex scenarios. Great strides have been made to realize WCS for the 24/7 monitoring of single users based on 3-tier architectures involving wearables, edge, and cloud systems. However, new requirements targeting cooperative multiuser scenarios demand radically new approaches, as promoted by the community-oriented WCS (CO-WCS). The COMMON-WEARS project is developing novel models and architectures for next-generation CO-WCS, defining a rigorous engineering methodology, with associated formal verification and simulation tools, to drive the development lifecycle of CO-WCS and creating a pilot in a real instrumented environment to support activities of surgical teams in operating rooms. COMMON-WEARS strives to move the research front by developing CO-WCS featuring multi-user collectives of smart wearables and body sensor networks (BSN), with applications, e.g., in domestic, urban, manufacturing, emergency, and working environments. This is achieved by combining aggregate computing, collective opportunistic IoT, machine learning, and WCS/BSN architectures. Herein, we discuss lessons learned from prototyping new concepts in collaborative wearable computing, in the contest of challenging use cases, including healthcare, emergency response and pandemic management, which are strongly impacted by smart wearables.
This paper presents a design method for the fuzzy state observer of spatial two-dimensional (2-D) nonlinear parabolic partial differential equation (PDE) systems under the utilization of a mobile sensor. Initially, a Takagi-Sugeno (T-S) fuzzy model is established, as it is capable of representing nonlinear systems. Subsequently, by employing the Lyapunov direct method, a fuzzy state observer and a guidance law of the mobile sensor are proposed to ensure the asymptotic stability of the state estimation error system. Ultimately, numerical simulations are conducted to demonstrate the viability of the developed approach.
Spatio-temporal systems which are described by partial differential equations widely exist in industrial applications such as the snap curing oven in chip manufacturing, the catalytic rod in chemical manufacturing, and the soft robots in special operations. The safe and reliable operation has gained increasing amounts of attention recently, which mainly focuses on the abnormality diagnosis issue. In this review, bearing the three indispensable questions about the concerned abnormality in mind, i.e., when, where and what (when an abnormality occurred? where the abnormal region is? what the magnitude is?), the abnormality detection, localization, and estimation techniques are introduced sequentially. Moreover, in each part, both classical literature and recent advances are introduced. Furthermore, some outlooks and future directions are presented. This review shall be beneficial in helping researchers obtain an overall view of this specific and other related research fields.
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
Industrial artificial intelligence (IAI) is an emerging enabler for the industry to transform toward digitalization and data-driven intelligence. Existing research works of IAI are mostly technology-driven, but technology alone is far from enough to realize the real impact of IAI on the industry. In this article, we go beyond technology and discuss IAI from a more comprehensive perspective in terms of its building blocks, including management, technology, and human. Those building blocks interact with each other and jointly enable IAI adoption in industrial applications. We discuss each of them as well as their synergy, considering various realistic aspects both technical and nontechnical. Our work is unique and has direct impact on the industry to well understand and embrace IAI for successful industry transformation.
This article is devoted to fixed-time (FXT) output optimization for heterogeneous linear multiagent systems (MASs) by proposing distributed piecewise protocols. By introducing an auxiliary variable for each agent, a state-based control scheme is designed and a distributed piecewise algorithm is developed for each auxiliary state. It is revealed that the FXT output optimization is successfully solved through three steps: accurately reaching the auxiliary state, realizing local optimization, and achieving global output optimization. To avoid the inaccessibility dilemma of state information, an output-based control protocol is further developed and an output-based distributed algorithm is proposed to realize output optimization in a fixed time. Finally, the proposed optimization algorithms are further illustrated via numerical results.
