Brownfield redevelopment represents a key strategy for sustainable urban transformation, yet mobility management in capacity-constrained industrial campuses remains insufficiently explored. This paper proposes a digital twin framework for corporate mobility management applied to the Cesana brownfield in Mlada Boleslav. The model integrates travel demand, behavioural adaptation, infrastructure capacity, and policy scenarios within a dynamic simulation environment. Analyses indicate that coordinated multimo dal strategies and incentive mechanisms can significantly reduce car dependency without infrastructure expansion. The framework demonstrates potential as a transferable decision-support tool for mobility planning in industrial brownfield contexts.
The study evaluated chatbots designed as conversational assistants for drivers with the aim of reducing driver fatigue through appropriate and undemanding conversation. The introductory part included a questionnaire survey focused on identifying preferred topics of conversation while driving, which provided a basis for selecting the content focus of the experiment. This was followed by a subjective evaluation of selected chatbots using user experience metrics, focusing on their comprehensibility, naturalness, and ability to maintain attention without increasing mental load. The final part used the QFD method to assess the extent to which individual chatbot procedures and features met the set criteria. The output was a comparison of chatbots and a determination of their suitability for further extensive experiments.
Maintaining consistent product quality is a key challenge in automotive manufacturing, where high production volumes and product variability place significant demands on inspection processes. Industrial computer vision (ICV) offers an effective approach for automating visual quality control using modern image processing and deep learning techniques. This paper presents a case study of an components on a pre-assembly production line. The system integrates industrial cameras, edge processing devices, and neural network models trained on annotated production datasets. The paper describes the system architecture, dataset preparation, model training, and integration with production monitoring tools. The deployed system inspects several million components annually and demonstrates reliable defect detection performance under real manufacturing conditions. The study highlights the practical benefits of industrial computer vision for large-scale automotive quality control and outlines future development directions including digital twin integration and predictive analytics.
Road surface abrasions significantly contribute to vehicle collisions and mechanical failures worldwide. Traditional machine learning-based methods for road damage detection typically rely heavily on extensive manual annotations, making them costly, labour-intensive, and inefficient. To address this challenge, this paper proposes a label-efficient image processing framework based on self-supervised representation learning for road damage classification. Our approach integrates contrastive learning with a regularized redundancy reduction method, enabling the extraction of rich, discriminative features directly from unlabelled data. Contrastive learning separates positive and negative samples to learn robust feature representations, while a cross-correlation loss maximizes information content by minimizing redundancy. Regularization through variance and covariance loss terms ensures feature diversity and prevents informational collapse in the learned representations. Extensive evaluations in both in-domain and cross-domain scenarios demonstrate that our proposed method achieves superior performance compared to supervised techniques, even when trained with substantially fewer labelled samples. Thus, this work provides an effective, economical, and scalable solution to the critical challenges faced in automated road maintenance. The downstream task considered in this study is multi-class classification of road damage categories rather than binary damaged-versus-undamaged road detection.
Public transport systems face increasing pressure to improve reliability under constrained infrastructure and limited investment capacity. Small-scale bus priority interventions represent a cost-effective tool for improving operational performance, yet their implementation requires reliable identification and prioritisation of delay-generating locations. This paper presents a fully automated method for spatial identification and quantification of delay formation in bus transport systems based exclusively on high-resolution AVL data. The proposed approach reconstructs vehicle trajectories using a high spatial resolution vectorized road network and map-matching, enabling continuous delay estimation along the entire route rather than only at stops. Referential travel times are derived empirically from historical data using a percentile-based approach, allowing delay quantification independently of static timetables and accommodating heterogeneous operating conditions. The method supports aggregation across multiple trips, lines, and corridors, providing a system-wide view of delay accumulation and its infrastructurerelated causes. The methodology is demonstrated on regional bus services. An experimental evaluation of machine learning models as substitutes for referential journeys indicates that, given the available data structure, AI-based approaches fail to achieve meaningful predictive performance and cannot reliably replace the proposed statistical reference. The presented method offers a scalable and robust decision-support tool for prioritizing bus priority interventions and improving public transport reliability using operational data already available to most transport authorities.
