This workshop will showcase research from the Future Roads programme, highlighting practical innovations across five key themes: Digital Twins, Sustainability, Smart Materials, Automation & Robotics, and Data Science.
Participants will gain insights into:
- How digital twins are evolving to become more user-friendly and integrated with real-time data.
- New materials and technologies that enable greener, smarter, and more resilient road infrastructure.
- The role of AI and robotics in improving safety, efficiency, and adaptability on construction sites and in traffic systems.
- Data-driven tools for predictive maintenance, flood resilience planning, and automated road condition assessment.
FUTUREROADS Programme Workshop
Efficient Reservoir Computing-based Control for Stabilising Mixed-Autonomy Traffic
Dr. Kai-Fung Chu
Traffic shock waves and the resulting congestion pose significant challenges to travel time and fuel efficiency. Autonomous vehicles (AVs) have emerged as a potential solution to alleviate these issues by intelligently controlling their speed. However, existing reinforcement learning (RL) methods for AV control are often inefficient, requiring extensive training time and data. This paper presents a novel reservoir computing (RC)-based control approach for longitudinal control of an AV operating within a vehicle fleet. The goal is to mitigate traffic shock waves by controlling an AV in mixed-autonomy traffic, achieving comparable performance to state-of-the-art RL methods in an efficient and adaptive way. Unlike RL methods that require substantial training of large data from various traffic conditions, our proposed approach utilize the intrinsic memory and dynamics of the traffic, supporting it to demonstrate superior data and training efficiency, and adaptability to varying traffic conditions. Simulation results on different road structures and configurations validate the method's effectiveness, showing comparable performance in mean speed, speed stability, traffic flow, and fuel efficiency. This demonstrates the advantages of RC-based control in terms of rapid training and data efficiency, making it a promising approach for efficient AV control in mixed-autonomy traffic. It also contributes to the understanding of leveraging system dynamics and physical intelligence rather than relying solely on computational solutions, leading to a significant social impact in real-world traffic and other applications where extensive data collection is impractical.
PaveMove: A novel theoretical model for TSD tests of asphalt pavements
Dr. Zhaojie Sun
This presentation introduces a self-developed theoretical model for Traffic Speed Deflectometer (TSD) tests of asphalt pavements. The detailed features, functions, and performance of PaveMove will be covered. The PaveMove can be used as a solver in the interpretation process of TSD measurements, which can produce structural health information for a digital twin of roads.
A graph-centric framework for digital twinning the built environment
Dr. Junxiang Zhu
This presentation is about graph generation, graph query, and graph-based knowledge discovery for digital twinning, based on graph-based representation of asset data.
Enabling Infrastructure Subsystems Integration via Agentic Ontology Engineering
Dr. Mengtian Yin
Semantic harmonisation of heterogeneous infrastructure information systems is foundational to integrating data and systems for infrastructure asset management (IAM). However, manual knowledge engineering over complex organisational databases is labour-intensive. This paper presents an agentic ontology engineering method that uses large language models (LLMs) within a multi-agent framework to automatically extract knowledge and model ontologies from heterogeneous IAM databases. The system progressively generates classes, hierarchies, relations and logical restrictions, while agents align cross-system concepts and detect contradictions to ensure consistency. Evaluated on four real-world IAM schema sets, the method achieves 92.1% entity recall against a ground truth ontology, 90.91% entity classification accuracy, and 83.18% relation classification accuracy. Compared with a single-LLM baseline, the multi-agent approach yields higher quality ontologies with fewer constraint violations. The resulting IAM ontology supports interoperability across diverse subsystems, demonstrating that agentic, LLM-driven modelling can significantly reduce effort while improving fidelity for infrastructure information management.
Road surface monitoring based on physical computing methods
Dr. Xiang Wang
Civilian vehicles installed with sensors have been used for road surface condition monitoring. The collected data are generally analysed with algorithms. Vehicles are dynamic systems which interact with the road via the tyres, so the collected dynamics features include short-term information of the system. Based on it, if another physical dynamic system is used to analyse the collected signals, the signals can be processed according to the dynamic characteristics of the physical system to obtain information such as road surface conditions. Previous studies have shown that using the inherent dynamics of an optoelectronic system can realise road surface condition analyses based on IMU data from a vehicle.
Towards Circular Pavements: Advancing Low-Carbon Cementitious Materials and Recycled Asphalt for a Net-Zero Future
Dr. Abbas Solouki
This talk explores innovative approaches to achieving circularity and carbon reduction in pavement engineering. It presents recent advances in the development and implementation of low-carbon cementitious materials such as geopolymers and limestone calcined clay cement (LC3) combined with recycled asphalt pavements (RAP) to design sustainable and high-performance road structures. The presentation will highlight laboratory results, material design strategies, and pathways for field application, emphasizing how these solutions contribute to the transition toward net-zero infrastructure.
Acknowledgements
This programme is funded by the EU Horizon 2020 under grant agreement No. 101034337.
Loading... 
