Project Summary

SYSTRA has developed an artificial intelligence-enabled tool to help reduce and mitigate the impact of road collisions at the design stage of new infrastructure. Designers are advised on how best to make roads and junctions safer as drawings take shape. Machine learning algorithms applied to existing data sources provide this insight by quantifying and qualifying road safety risks associated with existing and proposed highway designs.

Project Achievements

Taken proof of concept to a minimum value product served in a website application. Our proof of concept was a machine learning model that could classify physical environments using geospatial data. Using our physical environment classification model, we generated environment classifications for all collisions recorded in STATS19. These classifications are used as part of an application whereby the user can select any location on the Great Britain road network and be returned a specified number of collisions at similar physical environment. This provides some inference as to the type and severity of road collisions anticipated for the given location.

Conclusions

We would consider our TRIG project a success. We achieved one of our two key deliverables which enabled a marketable product. We were able to talk to potential stakeholders during the development process to gauge alignment with their requirements. This process led us to divert away from our second key deliverable as it became obvious that such functionality was not immediately demanded by our envisaged market. This engagement narrowed our scope and aided in delivering a high-quality product at the conclusion of the project.

Next Steps

We have a minimum value product which we are showcasing to our envisaged stakeholders – local authorities and highway agencies. We are confident that we will achieve a small degree of sales based on the existing capabilities of the application that we will use to enhance marketability to other stakeholders through further product development.