In order to address the limitations of the conventional traffic signal control methods with regard to adapting to dynamic traffic scenarios and the challenges associated with multi-objective collaborative optimization, this study proposes a Firefly-guided Hybrid Deep Q-Network (FH-DQN) algorithm for optimizing the intersection signal timing. The proposed framework synergistically integrates firefly algorithm-based swarm intelligence with deep reinforcement learning, employing a brightness-driven hierarchical exploration strategy for enhancing the action selection efficiency. A multi-objective dynamic reward function incorporating vehicle delay, queue length, carbon emissions, and traffic throughput is developed in order to achieve a balanced optimization of traffic efficiency and environmental sustainability. The algorithm architecture includes the following innovative components: a dual-network collaborative structure that enhances the vehicle responsiveness to sudden congestion at individual intersections, an adaptive brightness update rule derived from the firefly algorithm-based optimization principles and the multi-objective dynamic reward function which achieves the dynamic adjustment of the intersection traffic signals. The extensive experiments conducted on the Simulation of Urban Mobility (SUMO) platform demonstrate that the FH-DQN algorithm achieves a superior performance in comparison with the fixed-time and conventional DQN approaches in typical cross-intersection scenarios. Specifically, the average queue waiting time for vehicles when applying the proposed method is reduced by 26.09% in comparison with the DQN-based approach. The ablation experiment confirms the individual contributions of the dynamic reward function and global network components to enhancing the overall performance of the FH-DQN algorithm. To sum up, this study provides a novel framework for collaborative traffic signal optimization in complex urban road networks, significantly improving both the adaptive capability and multi-objective optimization performance of intelligent transportation systems. Future work could focus on adapting the algorithm to metropolitan-scale networks and on integrating multimodal traffic data for enhancing the operational robustness of the proposed model.
This study explores the integration of AI-generated synthetic data into deep learning pipelines for an automated waste detection and classification in industrial recycling systems. Based on the YOLOv12 object detection framework and the publicly available WaRP dataset, the proposed model was trained for recognizing six major waste categories, including bottles, cans, cardboard, and glass. In order to address data scarcity and class imbalance, up to 10,000 synthetic images were generated by using ChatGPT's image generation capabilities, compositing realistic waste objects onto clean conveyor-belt backgrounds. The experimental results demonstrated that augmenting the employed dataset with synthetic samples improved the proposed model`s detection and classification performance, which is proven by an increase from 0.593 (for the baseline model) to 0.622 for the mAP@50 metric and from 0.466 to 0.504 for the mAP@50:95 metric. The best results were achieved for the model augmented with 5,000 synthetic images, after which there was no further improvement in the performance of the employed model. These findings highlight the fact that high-quality synthetic data can effectively enhance deep learning models in waste sorting applications, reducing the dependence on extensive manual data collection. However, for further improvements it would be necessary to enhance the asset realism and diversity rather than simply increasing the dataset size. To sum up, the proposed approach underscores the potential of combining generative AI and computer vision for accelerating industrial automation.
Proportion-Integral-Derivative (PID) controllers are widely used for process control across various industries, including the chemical industry. However, the traditional one-degree-of-freedom PID controller struggles to balance a good setpoint tracking and disturbance rejection. This paper proposes two novel two-degree-of-freedom nonlinear PID (TDOF NPID) controllers. As such, this approach combines a nonlinear PID controller with its specific strengths with an independent linear compensator, which can be either a P-type or a PD-type compensator. A hybrid genetic algorithm is also employed for optimizing the parameters of the TDOF nonlinear PID controllers, considering both the setpoint tracking and disturbance rejection performance, with the goal of minimizing the integral of absolute error criterion. The performance of the proposed controllers is evaluated by benchmarking them against the traditional PID controller for three processes with varying orders and one nonlinear system.
In natural environments, interception represents a predatory behavior as the hunter anticipates the prey's path and moves toward a future position of the target rather than following it directly. This approach generally reduces both the time required to capture the prey and the associated energy consumption. Based on this concept, this paper introduces an interception-based controller-independent approach for tracking the trajectory of a differential-drive mobile robot. The proposed method is implemented and a fair comparison is carried out with two classical approaches and a recent hybrid prescribed-time controller (HPTC). Furthermore, the gains of the controllers are tuned using a Particle Swarm Optimization algorithm. The simulation results demonstrate that the proposed interception-based strategy provides an improved tracking accuracy and a faster convergence in comparison with the classical trajectory tracking methods.
This study presents a multi-objective optimization model for drone path planning that simultaneously considers path length, flight safety, and path diversity. The drone flight environment is first modeled, followed by the construction of an optimization model based on the path length and the drone flight-related threat level. An improved Particle Swarm Optimization (PSO) algorithm is then proposed, incorporating a Sugeno function for dynamically adjusting nonlinear inertia weights and learning factors, thereby enhancing the proposed model's global search capability and drone path planning accuracy. The experimental results for two test scenarios (Maps A and B) demonstrate that the proposed algorithm identifies the optimal drone flight paths after 68 and 75 iterations, respectively, with average path lengths of 1.52 km and 1.65 km. These findings show that the enhanced PSO algorithm provides an effective and reliable solution for drone path planning, with broader implications for algorithmic optimization in related fields.
This paper proposes a hybrid multi-criteria decision-making (MCDM) framework for selecting the optimal AI algorithms in the context of real-time infrared signal detection systems. Five performance criteria were considered, namely the processing speed, detection accuracy, segmentation efficiency, noise robustness and energy efficiency, reflecting the requirements of real-time image processing and embedded computer vision systems. This framework integrates the SWARA method for expert-based criteria weighting with Net Worth Analysis (NWA) for algorithm ranking, enabling a transparent and systematic evaluation. The experimental results show that the Fast R-CNN algorithm achieves the highest overall performance, while algorithms such as EfficientDet obtain lower scores and require further refinement to be effectively used in real-time infrared signal detection applications. To sum up, the proposed method addresses the current lack of structured decision-support tools for selecting among various AI-based infrared signal detection models under operational constraints. The research findings provide actionable guidance for researchers and practitioners developing embedded AI, surveillance and automated monitoring systems.
Multi-Criteria Decision Making (MCDM) methods are effective in solving complex selection problems with numerous conflicting criteria. However, the traditional methods may not succeed in structurally modeling the decisionmaker's psychological dynamics related to satisfaction and dissatisfaction. This study proposes a novel MCDM method inspired by Herzberg's Motivation-Hygiene Theory, which categorizes the employed criteria into two types: hygiene factors, whose absence causes dissatisfaction, and motivation factors, whose presence enhances satisfaction: the Dual-Factor Decision Making (DFDM) method. This method incorporates a strict threshold-based veto mechanism for hygiene factors and a dynamic contribution analysis for motivation factors. This approach distinguishes itself from the traditional compensatory methods (e.g. TOPSIS, AHP) by completely eliminating the alternatives that violate the threshold values, a feature which is reminiscent of the veto mechanism in the context of the ELECTRE methods. Furthermore, if a hygiene factor exceeds its threshold by a specified margin (delta), it will act as a motivation factor, enabling a dynamic evaluation. The applicability of the proposed method is demonstrated through a supplier selection case study. The obtained results indicate that the DFDM method can effectively identify and exclude the alternatives that violate critical constraints, such as budget limits, while providing a ranked list of feasible options. The DFDM method stands out as a human-centered decision support tool, making it particularly useful in domains with critical thresholds where psychological satisfaction is also important, such as public procurement, supply chain management, and human resources.
The employee performance evaluation and optimization is a key objective for modern organizations, where decision making in human resource management must be supported by robust analytical methods. This study proposes a modeling framework based on the integration of Self-Organizing Maps (SOMs) and kinematic models (kinMods) for the analysis and dynamic simulation of the employee performance under different managerial scenarios. SOMs are employed for classifying the employees according to their skills, performance, and potential, thereby identifying latent patterns in the team structure. In addition, the kinMod enables the simulation of managerial scenarios-such as promotions, restructuring, or the departure of key employees-and the assessment of their impact on organizational cohesion and efficiency over time. The experimental results, illustrated by the correlation analysis, employee clustering outcomes, and dynamic simulations, confirm the usefulness of the proposed methodology for reducing subjectivity in human resources assessment and providing an objective decision support tool. The integration of SOMs and kinMods thus provides a transferable methodological framework with potential applications in HR analytics and organizational management.
Despite the availability of numerous models, instruments, and theoretical frameworks on user experience (UX) evaluation, significant challenges remain in understanding UX from a holistic perspective, both theoretically and methodologically. This work contributes to the UX research by integrating contextual perspectives with a specific focus on mobile applications. This study aims to evaluate user experience and to identify the key factors influencing efficiency, satisfaction, and the long-term acceptance of mobile applications5. In order to address these challenges, this study proposes a novel longitudinal and context-aware methodological framework that combines subjective self-reported measures with objective and behavioral data. This framework7 is empirically validated through a three-phase longitudinal field study conducted under real-life usage conditions, by employing the Moodle mobile application.
The vibration control effectiveness for a vehicle using a controlled suspension system will outperform that of traditional passive suspension systems. The latter are limited with regard to their ability to adjust to varying road conditions, whereas the semi-active systems provide ahigher flexibility and efficiency. This paper introduces anew approach for improving the ride comfort with regard to the semi-active suspension systems of vehicles by employing the Deep Deterministic Policy Gradient (DDPG)-based algorithm. The main objective of this research is to optimize the vertical control of a semi-active suspension system by using deep reinforcement learning (DRL), providing a dynamic, real-time adaptive solution to varying road conditions. By employing the DDPG-based algorithm, this paper addresses challenges such as continuous state spaces and action execution in semi-active suspension systems. The semi-active suspension system is modeled dynamically and its responses are evaluated across various road surfaces, including uneven and bump road conditions. The simulation results show that the DDPG-based suspension system significantly improves the suspension performance by reducing the vertical body acceleration, the suspension deflection, and the overall passenger discomfort in comparison with the passive suspension systems. The proposed approach features an excellent adaptability and efficiency in vehicle suspension control, highlighting the potential of AI in optimizing vehicle suspension systems for an enhanced ride comfort and stability.
