This dataset contains over 7 million high-quality annotations of both railway-specific and general perception objects, captured under varying operational scenarios, to serve as a valuable resource for advancing AI-driven environment monitoring in the railway domain.
Abstract
Reliable environment monitoring is essential for the safe and efficient operation of automated railway systems, covering all Grades of Automation (GoA), from partially automated (GoA2) to fully automated operation (GoA4). Artificial Intelligence (AI) plays a central role in enabling these systems to detect, classify, and react to potential hazards in real time. The development of such AI-based perception systems requires large volumes of accurately annotated data for training and validation. Within the Digitale Schiene Deutschland (DSD) program, DB InfraGO AG and understandAI GmbH have developed a comprehensive multi- sensor dataset tailored to the needs of railway environment perception. This dataset contains over 7 million high-quality annotations of both railway-specific and general perception objects, captured under varying operational scenarios. The finalized dataset can now be requested at the DB InfraGO AG and serve as a valuable resource for advancing AI-driven environment monitoring in the railway domain.
This review paper argues that Vision-Language Models (VLMs), a paradigm whose rapid maturation is evidenced by recent comprehensive surveys offer a transformative solution for rail asset management, and provides a focused overview of the limitations of current CV systems.
Ashley Varghese, Mohammadjavad Ghorbanalivaki, Gunho Sohn· The International Archives o...· 0 citations
This framework proves an autonomous decision-making system that organically links inspection data with maintenance regulations by transforming static, manual-labor-centered maintenance workflows into intelligent automated models and increases the efficiency of railway infrastructure management while providing a scalable technical foundation for overall asset management of future smart-city infrastructure.
Minjae Jeon, Yong-Gyun Kim, Seok-Han Kim· Smart Cities· 0 citations
As the requirements for railway operators to offer a safe and economically efficient transportation service have become extremely difficult to met due to the size and complexity of the rail network, vehicle-based track monitoring and fault detection have gained relevance as a way to abate maintenance-related costs via condition-based and predictive approaches. In this work, a cost-effective, permit-free vehicle-based track monitoring system implemented on a regular in-service vehicle is presented. The designed system makes use of inertial sensors mounted on the vehicle’s bogie and car body, as well as positioning technology based on global navigation satellite system (GNSS) to collect monitoring data for detection and spatial localization of track defects. The capabilities of the developed system are exemplified by means of a case study related to a track section with a temporary speed restriction (TSR)—a scenario where traditional acceleration-based detection often fails due to reduced vehicle speed. Here, it is demonstrated that the proposed approach, combining statistical and time-frequency analysis (e.g., wavelets), can effectively lead to the detection and localization of anomalies on the track using onboard inertial measurements, even under reduced vehicle speed.
Héctor A. Fernández-Bobadilla, R. Frolow, Laura T. Rodríguez-Bayona et al.· Railway Engineering Science· 0 citations
This study proposes an AI-enabled autonomous drone framework for infrastructure inspection that integrates intelligent flight planning, automated data collection, computer vision-based defect detection, and condition assessment that enhances inspection accuracy, operational safety, and scalability compared to traditional methods.
Yuki Nakamura· International Journal of Mod...· 0 citations
Vehicles that have intelligent perception systems are the first in the new wave of innovation that is geared towards enhancing transportation safety, sustainability, and efficiency. Such systems are a collection of sensors, artificial intelligence (AI), machine learning (ML), and real-time data processing to allow vehicles to sense and understand their surroundings with a great level of accuracy. In this review, we discuss the use of intelligent perception technologies in reducing environmental hazards and leveraging resources to the fullest. It discusses important sensor technologies, such as cameras, Light Detection and Ranging (LiDAR), radar, and ultrasonic sensors, as well as AI-based decision-making systems that enable vehicles to respond to changing driving conditions. Intelligent perception systems can help reduce vehicle emissions and minimize the environmental effects of cars by optimizing driving behaviors and consuming less fuel. The significance of optimizing resources is also addressed in the review, with references to the applications of intelligent routing, eco-driving, and autonomous vehicle systems. Although these technologies are promising, there are challenges such as the integration of the technologies, issues related to sensor fusion, data privacy, and infrastructure upgrade that are required. Also, new technologies like 5G connectivity, edge computing, and sensor technology will develop further to add additional features to the intelligent perception systems. The article ends by highlighting that such technologies can be transformative in the pursuit of sustainable, efficient, and safe forms of transportation.
Min Li, Qiaosheng Bo, Junrong Huang et al.· Journal of Environmental &am...· 0 citations
Since the technology of autonomous driving is gaining momentum in its implementation in real-life conditions with multifaceted and complicated road conditions, the weaknesses of single sensors in the context of sensing accuracy, stability, and adaptability to the environment become more evident. To improve the robustness and security of autonomous driving systems in compound environments, in this paper, the research on the multi-sensor fusion technology is put into the limelight and the value of such technology applied in autonomous driving perception systems are evaluated. Thereafter, a comparison of the perception properties, benefits, and deficits of cameras and lidar is made systematically and at the data level, the fundamental patterns of multi-sensor integration are divided into three levels, namely the feature-level, sensor-level and decision-level. This study shows that when multi-sensor fusion strategies are rationally designed, the constraints of a single sensor used to sample the environment can be addressed, and this strategy plays an important role in increasing the resilience of the system as well as its ability to understand its surrounding. The discussion made in this paper offers a useful source of information when it comes to the design and implementation of multi-sensors fusion systems within the engineering field.
Zhixiang Xu· MATEC Web of Conferences· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
