skip to content

Digital Roads of the Future


Title: Zero waste geopolymer pavements

Fellow: Dr Abbas Solouki 

Theme: Smart Materials

Abstract: The non-stop growth of highway construction and its developments could hugely affect the global supply of natural resources. Recycling waste materials such as reclaimed asphalt pavement (RAP) and recycled concrete aggregates (RCA) from pavements could mitigate this issue and provide zero-waste paving solutions. However, utilization of such material is still very limited, where 8.4% by weight of the RAP in Europe is stockpiled, and up to 20% is down-recycled as aggregates in unbound layers. The research aims at recycling the most pertinent material such as RAP, RCA and local waste into pavements from a multi-scale perspective. Fundamental project details such as available resources, materials, data, and equipment, potential waste streams and possible trial site locations will be identified through literature review and discussion with industrial partners. Producing High-RAP will be investigated by implementing performance-based methodologies rather than conventional methods. The incorporation of RAP and RCA into in-house produced geopolymer binders and its on-site implementation will be evaluated. Extensive laboratorial and large-scale testing will be 
conducted prior to Field trials. On-site trials will be planned and carried out by the help of the industry partners and will be monitored using radio frequency identification (RFID) sensors. The obtained data will be then utilized to optimize pavement design, promoting future application of such pavements. Life-cycle analysis will provide insight on the environmental, economic and CO2 footprint of the project. The outcomes of the study are expected to improve the understanding of geopolymer pavements and provide a field ready technology for potential large-scale adoption.

Title: A multi-agent system for heavy machinery operation through context aware sensor fusion

Fellow: Dr Chapa Hewa Pelendage

Theme: Automation & Robotics

Abstract: Heavy machinery poses a severe risk to humans in proximity. Despite specialized safety training, less experienced workers are considerably more likely to be involved in construction site related accidents. Hence the applicant would like to suggest a multi-agent system (MAS) comprised of heavy machinery, operators and mobile devices to monitor the surrounding workplace and assess potential hazards and risks while coordinating tasks within the system. The aim of this proposal is to improve the situation awareness of heavy machinery operators using sensory feedback and avoid potential risks. This will further make relevant information from the unreachable areas within the environment accessible for operators. To completely cover complex areas like construction sites with sensors, mobile sensing 
devices (MSD) will be considered. This proposal should be implemented in 3 stages: Firstly, the applicant will identify the requirements of construction workers and possible hazards to design a human machine interface (HMI). Operators will use this to interact with agents in the workspace. Secondly, an adaptive algorithm which allows the agents to explore the environment based on current operations will be developed. Thirdly the applicant will explore ways to improve the model such as using vision, acoustic and haptic feedback. The system will be implemented in the field in the second and third stages to increase the adaptive control and experience considering possible scenarios as much as possible. The proposed concept has the potential to make a considerable improvement to construction site safety, creating a high social and technical impact with multidisciplinary collaborations.

Title: Control and implications of mixed autonomous vehicle-infrastructure in a heterogeneous multi-agent system framework

Fellow: Dr Kai-Fung Chu

Theme: Automation & Robotics

Abstract: Connected autonomous vehicles (CAVs) will be prevalent given the technology maturity and government promotion. Transportation system administrators and constructors should be prepared to leverage the controllability and potential of CAVs when it gradually permeates the roads in the near future. Current studies for CAVs usually consider either the control of a single CAV with human-driven vehicles (HVs), or collectively control among CAVs only. Moreover, their interactions with intelligent road infrastructure are investigated separately, which oversimplifies the heterogeneity 
among CAVs, HVs, and intelligent road infrastructure in the near future transportation system. Numerous challenges of heterogeneous multi-agent transportation systems, such as their interactions, partial controllability, and implications are not addressed. This project will study the control and implication of the heterogeneous multi-agent transportation system mixed with CAVs, HVs, dynamic reversible lanes, and intelligent traffic lights. First, with the awareness of the high spatial-temporal resolution and real-time characteristics of transportation systems, an efficient heterogeneous data fusion and multi-agent modeling framework will be developed. Second, optimal control policies for the heterogeneous multi-agent transportation system satisfying the safety requirements will be developed. Third, the opportunities and barriers to practical implementation and the implications for society and governance will be analyzed. Through this project, we will push forward the technologies and insights of CAVs at dynamic reversible lanes and intelligent intersections, and move towards a safer and more efficient system whereby CAVs can be highly leveraged.

Title: Development of a vision-based method for high-quality traffic data collection in support of building trustworthy digital twins of road networks 

Fellow: Dr Linjun Lu

Theme: Digital Twins

Abstract: Traffic data is one of the vital components involved in the digital twins of road networks, which can be used to enhance the performance of road condition assessment and prediction and improve the intelligence of road transportation systems. With the advances in electro-optical technologies and computational power, computer vision-based technology has been developed as a cost-efficient tool to assist in building digital twins as it can provide both rich semantic and geometric information. Nevertheless, when it comes to traffic data collection, there are still some shared hurdles faced by the existing vision-based methods to achieve high data collection performance. This is primarily due to the practical constraints including occlusion, shadow, and irregularity of vehicles that prevent the existing methods from 
accurately extracting measurements about the vehicles. This project is expected to make contributions to the body of road digital twin knowledge by developing a new vision-based method that allows for reliable and accurate measurement of dynamic road user information. This information will be amenable to applications that require high-quality input of traffic scene data including building a trustworthy digital twin of road infrastructure.

Title: Highway semantic web creation, distillation, and data exchange for unifying Digital Twin data standards.

Fellow: Dr Mengtian Yin

Theme: Digital Twins

Abstract: A digital twin (DT) for highway infrastructure has potential to integrate fragmented asset and operational data for a comprehensive management, analysis, and augmentation. However, there is a lack of unified data standard of DT to clarify the information requirements and how data is exchanged between different sources. To fill this gap, this study aims to develop an ontology-based framework that contains the minimum data requirements of highway DT. To begin with, a large highway semantic web is created based on natural language processing (NLP)-based information extraction from a vast number of project documents and professional articles. Based on the highway semantic web and an interview with industry professionels, a requirement analysis is carried out to summarize the most fundamental and crucial data requirements for a highway DT. The highway ontology is then distilled to derive the core DT data model. Finally, to integrate the DT data standard with various data sources, a unified Graph Neural Network (GNN)-based link prediction method is proposed to connect entity nodes in heterogeneous graphs, which implies how different data models are related and exchanged. The output DT data model is assessed through real-world case studies to assess its feasibility, efficiency, and effectiveness.

Title: Digital Twins for Road Infrastructure Networks 

Fellow: Dr Munkhbaatar Buuveibaatar

Theme: Digital Twins

Abstract: Road networks are important to provide connectivity, transportation, and mobility for the people. Digital Twins (DT) for these road networks – highways are expected 
provide capabilities that will not be available when just having physical assets. This project aims to define the minimum data requirements (MDR) for a valuable DTs in general, and aims to seek answers for the following questions in particular: (1) What are the minimum data requirements (MDR) for a valuable DTs?; (2)What are unique (data) requirements for Highway DTs (HDT)?; (3) What are the standards for DTs?; (4) How to harmonize different DTs?; (5) What data can be shared between different infrastructures DTs (e.g. supply infrastructure)?; (6) How to integrate other data sources (e.g. railway, weather, environment data)?. To find out answers to these research questions, the following methodologies will be employed in common (but not limited to): (i) Literature Review (e.g. articles, documents); (ii) Overview of the current situation; (iii) Capturing user requirements and conducting required interviews; (iv) Inspecting available data and information (of existing Legacy systems); (v) Developing methodologies and verification; and (vi) Writing reports/papers and/or developing prototypes (if required). In addition, these methods may differ a bit depending on research questions. In addition, A CityGML geospatial open standard is intended to be used thoroughly to answer the questions. When the project achieves its goals, underlying factors for developing the minimum viable product of a DT will be defined, and it can be used for validating a product idea early in the product development lifecycle.

Title: Ribbons and fibres as a climate resilient solution for pavements

Fellow: Dr Quentin Félix Adam

Theme: Smart Materials

Abstract: Due to climate change, pavements in the UK will be exposed to colder and warmer air temperature levels in the winter and summer, respectively. This increase in air temperature amplitude is accompanied by safety concerns for the road network’s users, including the reduction of skid resistance and masking of road markings due to snow and ice in winter, and augmenting of aquaplaning risks due to bleeding and rutting in summer. Pavement materials are designed with current air temperature levels and are thus not proofed for the increase in air temperature amplitude. It is therefore of utmost importance to develop pavement materials that are capable of coping with the upcoming extreme temperature levels. This Fellowship is concerned with delivering a system consisting of embedded ribbons and fibres within pavements; this system is seen as a solution to the upcoming extreme temperature levels. The objectives are to extend my knowledge of heating ribbons to pavements with fibres, take to the next level the proposed system to mitigate bleeding and rutting, implement in full-scale the system to perform large-scale trials, and design a route to market for the commercialization of the system. The objectives will be fulfilled through laboratory work with the construction of a full-scale test section on which experimental work will be carried out. Modelling efforts for the automatization of the system will be performed. Implementing the suggested system in full scale is expected to be cost-effective, promote sustainability, prolong the lifetime of the pavement, and provide higher safety for users.

Title: Towards the development of a resilient highway digital twin: Requirements, Specifications and Standards

Fellow: Dr Varun Kumar Reja 

Theme: Digital Twins

Abstract: Digital twin technology holds high potential to assist in crucial decisions making. However, various challenges need to be addressed for their smooth adoption. Currently, no guidelines/standards exist on the minimum data requirements to create a valuable digital twin. These requirements are necessary at the product development stage to know the requirements for a digital twin's minimum viable product (MVP). These requirements will help validate the product idea in the development cycle and engage early-adopter customers for feedback. The key objective of this research is to solve these challenges by identifying the unique requirements to create a valuable highway digital twin. For this, a qualitative study will be conducted by engaging with various experts among the stakeholders, reviewing the relevant literature, and brainstorming. Next, the type of shareable data and data sharing mechanism between different infrastructure digital twins will be explored for getting accurate insights with minimal sensing by exploring cross-dependency between the digital twin attributes. Next, integrating other data sources, including traffic, railway, weather, environment, and seismic data, can give potentially helpful information. For this, the use of a common data environment will be explored. These proposed solutions will be validated experimentally by generating a workflow for an actual highway project. 
Finally, the unique requirements and shareable attributes will be methodically recorded, and conditions for generating a standard for highway digital twins will be listed. Though this will be the first step towards 
generating a standard, this research will open doors to attract early customers to adopt this technology confidently.

Title: Data fusion and data structure design in creating and updating digital twins

Fellow: Dr Yuandong Pan

Theme: Digital Twins & Data Science

Abstract: In this research proposal, I propose the project of data fusion and data structure design that aims to solve the challenge of “how can a Highway Digital Twin evolve over time”. An enormous amount of data from different sensors is collected and needs to be processed in the Highway digital twin project over time. Images and point clouds are two commonly used raw data to represent the geometry of the facility. In the project, I would like to propose an automatic method which combines object detection in 2D images and 3D point clouds together, based on my previous research finding that 2D detection and detection have different strengths and can provide complementary information. 2D images and 3D point clouds are registered together, and the corresponding extracted information is also linked. In addition, data management plays an essential role in a Highway digital twin project because of the large data volume over time. Therefore, I also want to design a data structure for the Highway digital twin, which aims to help in the effective management of large amounts of data over time, including but not restricted to data storage, data update, change tracking, and data querying. The extracted geometric information from the proposed 2D-3D combination approach is stored as basic constituents in the designed data structure, which would allow storing and linking of other kinds of data over time.

Title: Highway intelligent traffic control system based on vehicle-road coordination and multi-agent technology

Fellow: Dr Yue Xie

Theme: Automation & Robotics

Abstract: Increasing traffic congestion around the world leads to a series of adverse effects on the public travel and the development of society, such as travel delay, vehicle fuel consumption and environmental pollution. In recent years, autonomous driving has become an increasingly practical technology leading to new challenges and opportunities for traffic management on highways. This proposal expects to generate new knowledge in highway traffic management by developing a novel multi-agent control system that adopts reinforcement learning and heuristic approaches. The system aims to achieve the global optimization of the highway region, alleviate traffic jams, reduce travel times, and then increase traffic management efficiency by control of traffic instructions and optimal travel time. The system can be applied in various scenarios, such as only autonomous vehicles on the highways, both autonomous vehicles and human-driven vehicles on the highways or only human-driven vehicles on the highways. Reinforcement learning has great potential as a tool in traffic instruments control, while the existing algorithms have some drawbacks due to the cooperative control between agents and heuristic approaches have been successfully applied to optimization problems as well as cooperative optimization. The outcomes of this proposal will produce an intelligent and partial controllability multi-agent system that provides significant social, economic, and environmental benefits through optimal control strategies and effective management schemes

Title: An innovative zonation-based machine-learning methodology for studying the interactive impacts of traffic, microclimate and natural hazards on pavement deterioration

Fellow: Dr Ze Zhou Wang

Theme: Smart Materials

Abstract: This data-driven project aims to provide new knowledge on the interactive impacts of traffic, microclimate and natural hazards on pavement deterioration. A combined historical pavement image 
dataset obtained from Google Street View and Cambridge University will be processed using deep-learning algorithms to obtain the characteristics of pavement deterioration. Next, through a close collaboration with partner organizations, a wealth of datasets that include (i) traffic and heavy vehicle volume, (ii) national temperature and precipitation data, (iii) microclimate, (iv) extreme climate events, and (v) pavement material-related data are processed. To effectively handle the data, I propose to classify the data into three levels: micro-scale, meso-scale and macro-scale data. An innovative zonation strategy will be established to divide the country into multiple macroscopic and mesoscopic zones. The concept of hierarchical machine-learning will be utilized to develop a first model for all the macroscopic zones, 
which will then be used as the foundation model for the second machine-learning model developed for all the mesoscopic zones. In this way, it will be viable to provide refined predictions considering regional/location-specific information, e.g., microclimate. Ultimately, the new knowledge and the developed machine-learning methodology will be output as a standalone software, which will be implemented in practice for both pavement engineers and policy makers in partner organizations. The socially desirable outcomes will be: (i) the frequency of maintenance work can be significantly reduced, contributing to achieving carbon neutrality in U.K., and (ii) the new knowledge can lead to informed decision-makings on future road planning and design.