skip to content

Digital Roads of the Future

 

 

The key data science “challenges” are outlined below. 

How can uncertainty quantification be done effectively at scale?

Statistical reasoning about physical engineering systems is essential in this FR endeavour; however, even simple uncertainty quantification problems can be very computationally costly. Cloud computing (e.g. Amazon EC2) is an immense resource for Data Science and is potentially an ideally suitable platform for the Machine learning/Bayesian inference algorithms that will be employed to quantify the uncertainty of insights drawn from the noisy and disparate sources of data within the Digital Twin. The pertinent questions are thus how can these algorithms be run effectively at scale, say on the cloud, avoid idling computing resources and deliver timely results via real-time computing budgets? Furthermore, how can these algorithms be designed to be elastic, i.e. seamlessly utilise new computing resources as they come available and be robust to unexpected drop-outs of existing computing resources?

Industry Sponsor: Ramboll

How data can improve safety during Highways Activities?

Highways activities, such as maintenance and replacement, require temporary traffic management. “Live Traffic” delivers real-time traffic updates, dynamically rerouting vehicles to avoid hold-ups. However, this procedure poses safety risks, such as vehicle incursions into road works. The frequency of Live Traffic is constantly growing, and consequently, the safety risks, since the focus of Transportation Authorities is more on updating existing infrastructure than developing a new one. The frequency increases even more due to the need for upgrading infrastructure to support electric and autonomous vehicles. The major progress of the last two decades in collecting and processing data, using artificial intelligence, smart materials, automation and robotics has the potential to improve safety during Highways Activities significantly. The research question is how we can take advantage of data collection and process to improve safety during Highways Activates.

Industry Sponsor: BAM Nuttall

How can we improve Trust in Data and Artificial Intelligence algorithms with highways applications?

Collected data can help us monitor and predict traffic, assets condition and weather. Consequently, there is a great potential to improve highway monitoring, maintenance, traffic delays, users’ comfort and safety. However, to what extent do we trust the data, i.e. its veracity and the results from the artificial intelligence algorithms? For instance, road users do not always follow instructions related to the path they need to follow to reach their destination. Moreover, road managers do not wholly trust automated techniques for asset condition monitoring, and they combine automated with manual monitoring. So, how can we define trust in data and artificial intelligence? How can we measure it? How can we improve it?

Industry Sponsor: Atkins

Contact for Data Science Theme

Potential applicants should contact the data science theme lead, Professor Sumeetpal S. Singh (sss40@cam.ac.uk), for any queries regarding these challenges.

For the cross-cutting challenges 4-6 applicants should also choose to contact Professor Ioannis Brilakis (ib340@cam.ac.uk).

More on Data Science