
Dynamic Mobile Sensing for Resilient Transport Infrastructure
Dynamic mobile sensing technologies enhance condition monitoring and ensure resilient transport infrastructure.
- A collaborative initiative led by the Singapore-ETH Centre and the Swiss Federal Institute of Technology in Zurich (ETH Zurich), in partnership with the National University of Singapore, Nanyang Technological University (NTU), the Land Transport Authority of Singapore (LTA), and Swiss Federal Railways (SBB), pioneers dynamic mobile sensing technologies to enhance condition monitoring and ensure resilient transport infrastructure.
- By integrating real-time sensing with AI-driven analysis, this project enables cost-effective and scalable inspections across roads, bridges and railway networks. The system uses robots, trains, bicycles, and cars to collect high-resolution data, identify vulnerabilities, and improve inspection efficiency.
- Field tests in Singapore and Switzerland have validated the effectiveness of this solution across diverse environments, potential cost reductions in maintenance across different scales and applications, including roads, bridges, and railway assets.
The Critical Role of Infrastructure Maintenance in Preventing Disruptions
In the densely populated and land-scarce city-state of Singapore transport networks are constantly under pressure.
In 2017, a major MRT breakdown at Joo Koon injured commuters, and in 2024, the longest disruption in Singapore’s East-West line history caused severe delays, impacting 2.1 million commuters over six days. These disruptions not only inconvenience passengers but also create a cascading effect, stressing alternative transport modes like buses and taxis, and putting additional pressure on emergency services. Such disruptions are becoming more frequent, occurring every 3-5 years, underscoring the high likelihood of such events and the need for continuous, proactive infrastructure monitoring.
This expensive lesson underscores the need for robust, advanced, and continuous monitoring of infrastructure(s) throughout their life cycle. Availability of such monitoring data (ideally in real-time) provides invaluable decision support to optimise operation and maintenance processes. This also guarantees the safety and resilience of critical transport assets.
Current State of Infrastructure Monitoring and Its Challenges
Common methods of infrastructure monitoring, such as visual inspection, or fixed sensor networks installed directly on bridges, roads, and railways, are referred to as fixed sensing for Structural Health Monitoring (SHM).
However, this classical approach faces several limitations. First, the deployment of fixed sensing systems is labour-intensive and time-consuming, meaning that only a small fraction of transport assets can be covered. With resources stretched thin, the majority of critical structures remains unmonitored and the potential risks unaddressed. .
In contrast, a new paradigm of monitoring solutions known as mobile sensing has emerged in recent years, facilitated by algorithmic and technological advances, including AI. Unlike fixed sensing, mobile sensing deploys sensors mounted on moving platforms - such as drones, cars, trains, and robots - to inspect infrastructure. This enables more efficient, and scalable inspections.
Mobile sensing introduces a proactive monitoring approach, allowing early detection of structural weaknesses and reducing the risk of failures. Mobile sensing platforms help transport authorities prioritise maintenance efforts more effectively, reducing disruptions and improving long-term transport resilience.
Transitioning to Mobile Sensing for Smarter Monitoring

Bridge Inspection
Researchers introduced a wheeled robot equipped with advanced sensors designed to capture bridge vibration signals and transmit data back to engineers for real-time monitoring of the structural health of the bridges. Integrated with 4G communication and Global Navigation Satellite System (GNSS) modules, this robot performs AI-assisted automated inspections based on remote commands.
This robot has already been successfully tested on three bridges in Singapore, with the support of NTU, NUS, and the LTA - a novel advancement in fast and automated bridge inspections.

Road Inspection
Even humble vehicles, such bicycles, can be leveraged when it comes to road condition monitoring. Researchers developed a compact sensor box, which can be easily attached to bicycles to measure road roughness through vibration data. Equipped with an accelerometer and GPS antenna, the device maps vibration patterns to specific locations. It was tested on various roads in Zurich and Belgium to validate road inspection models and later, in Singapore to assess park connectors, which are vital for both recreation and transport.
Turning Data into Action

The research team has created a Proof of Concept virtual platform to receive, visualise and process and real-time data transmitted from the abovementioned mobile sensing devices.
Using the park connector network as a simulated test case, the platform demonstrates how data from bicycles equipped with sensors can be streamed, analysed, and mapped. Future upgrades will integrate maintenance planning tools, providing policymakers and planners with actionable insights to optimise infrastructure upkeep.
Looking Ahead: Offering Tangible Solutions for Resilient Transport Infrastructures
When it comes to sustainable transport infrastructure, people often think of low-emission transport systems. However, maintaining the safety and functionality of existing infrastructure to minimise disruptions are just as essential. By extending the lifecycles of transport infrastructure and minimising disruptions, we ensure long-term resilience.
Yet, maintaining the functionality and safety of these structures is a growing challenge. Factors like aging, environmental wear, and heavy usage challenge its upkeep.
Railway Inspections and its Potential for Future Research

The ETH team has developed innovative methods to assess railway conditions using accelerometers installed on diagnostic and in-service trains. These sensors collect vibration data, which is analysed to detect rail roughness; a critical sign of wear and tear that could lead to disruptions if left unaddressed.
By studying vehicle-based monitoring data over a 10-year period, the team has demonstrated how these insights can be integrated into decision-making tools to help plan optimal maintenance schedules for railway infrastructure, even when facing uncertainties. Moving forward, the team aims to refine these methods further to evaluate the structural health of railway bridges.
The ETH team's innovative approach to railway inspections and its seamless integration into maintenance planning highlights the transformative potential of data-driven infrastructure management.
In this endeavour, the international, cross-disciplinary collaboration between the FRS and ETH research team builds on theoretical frameworks to devise practical methods, software, and hardware prototypes to enable mobile sensing for infrastructure assessment.
In particular, the development of the Dynamic Mobile Sensing Platform demonstrates their commitment to enhancing the resilience of these vital networks. This system empowers decision-makers with critical, real-time data, enabling informed maintenance decisions which better support the long-term safety and functionality of transport infrastructure in both Singapore and Switzerland.
These advances represent a significant step toward more resilient, well-maintained transport networks, ensuring reliable mobility for future generations.
Contributors
Principal Investigators: Prof. Eleni CHATZI (ETH Zurich, SEC), Asst. Prof. LAI Zhilu (HKUST(GZ)/HKUST, formerly SEC), Asst. Prof. FU Yuguang (NTU), Prof. KOH Chan Ghee (NUS), Prof. Andreas WIESER (ETH Zurich)
Researchers: Dr. JIAN Xudong (SEC), Dr. Charikleia STOURA (ETH Zurich), Kiran BACSA (ETH Zurich, SEC), LIU Wei (NUS, SEC), TONG Yuzhou (formerly NTU & SEC), CUI Shuaiwen (NTU), Dr. Matej VARGA (ETH Zurich), Dr. LIANG Huangbin (SEC)
Further readings
[1] Jian, X., Lai, Z., Bacsa, K., Fu, Y., Koh, C.G., Sun, L., Wieser, A. and Chatzi, E., 2024. A Robotic Automated Solution for Operational Modal Analysis of Bridges with High-Resolution Mode Shape Recovery. Journal of Structural Engineering, 150(8), p.04024081.
[2] Stoura, C.D., Dertimanis, V.K., Hoelzl, C., Kossmann, C., Cigada, A. and Chatzi, E.N., 2023. A Model‐Based Bayesian Inference Approach for On‐Board Monitoring of Rail Roughness Profiles: Application on Field Measurement Data of the Swiss Federal Railways Network. Structural Control and Health Monitoring, 2023(1), p.8855542.
[3] Arcieri, G., Hoelzl, C., Schwery, O., Straub, D., Papakonstantinou, K.G. and Chatzi, E., 2023. Bridging POMDPs and Bayesian decision making for robust maintenance planning under model uncertainty: An application to railway systems. Reliability Engineering & System Safety, 239, p.109496.
[4] Liu, W., Lai, Z., Stoura, C.D., Bacsa, K. and Chatzi, E., 2022. Model-based Unknown Input Estimation via Partially Observable Markov Decision Processes.