School of Mathematics, Statistics and Physics

News Item

Using safety cameras and statistics for keeping our roads safe

We have developed software that provides accurate collision hotspot predictions

Writing for Public Sector Executive: PSE, Dr Neil Thorpe, Dr Lee Fawcett, Matt Linsley and Brett Cherry discuss sustainably and efficiently improving road safety in the UK.

In the UK, over 80% of passenger miles are by car, van or taxi. But around five people die every day from traffic collisions on Britain’s roads. In 2017, if all collisions were prevented, it would have saved the country an estimated £35bn, or 1.4% of GDP.

Great progress has been made in reducing road fatalities, but it has stalled in recent years. One way to re-establish the long-term downward trend is by preventing speeding. Safety cameras have played a major role in this since the 2000s.

Nearly half of all cars on Britain’s motorways exceed the speed limit. Speeding is also a challenge for road safety in our towns and cities.

Faced with a limited budget, when and where should authorities place safety cameras to make the most difference? GIS and other technologies help identify where so-called ‘collision hotspots’ occur. But a statistical approach can predict where they might occur in the future to help prevent them.

Predicting collision hotspots

Our researchers have developed software that integrates collision hotspot prediction with existing traffic monitoring and management systems.

The work is co-funded by the Northumbria Safer Roads Initiative (NSRI), the Engineering and Physical Sciences Research Council (EPSRC), PTV Group and Newcastle University.

We worked with a wide range of stakeholders, including highway operations and management organisations, law enforcement agencies, and local and regional transport planning authorities.

Making the most of data

The Newcastle University approach uses recent and historical records of collision counts to make predictions of traffic collisions. It puts more emphasis on recent observations for making predictions. It incorporates multiple crash counts from different points in time. The accuracy of its estimates is better than other approaches for collision hotspot prediction.

First, it looks at data on potential collision hotspots from across the road network to estimate the underlying average collision rate. It then combines them with observed collisions and local trends.

It is proactive rather than reactive, allowing practitioners to act and implement road safety mechanisms before collisions occur. This potentially decreases the number of crashes.

Better return on investment

Road safety budgets are continually under pressure. Placing safety cameras effectively will have the greatest return on investment. We can’t account for how many crashes would decrease regardless of intervention. Thus, it’s difficult to say with confidence if and by how much safety cameras reduce the number of crashes. The Newcastle approach gives stakeholders confidence in planning decisions for road safety.

In using cameras to deter speed, you want to make every penny count in preventing serious injuries and fatalities. The software helps make this possible. It enables authorities to fulfil their responsibility in keeping roads safe.

Using statistics to save lives

No other software tool like the one developed at Newcastle University has been in use for predicting collisions. It has potential to do so in real-time. Its use has resulted in lower casualty rates overall.

In the UK, North Yorkshire County Council and North Yorkshire Police were successful in assessing the performance of their safety cameras. The project covered 22 traffic sites from 2011-14. There were 46 casualties recorded in the period before intervention, and 33 after. The Newcastle analysis revealed that a reduction in eight casualties was attributed to the mobile safety cameras. The remaining reduction of five was due to other effects.

North Yorkshire Police have expanded their fleet of mobile road safety cameras to reduce the number of collisions, deaths and serious injuries further on the region’s roads.

Northumberland and Tyne & Wear used the results to predict future collisions and casualty rates. This made a difference for both regions. It helped them predict actual collision hotspots.

The NSRI is a partnership between Northumbria Police and six local authorities to reduce the number of fatalities and injuries on the roads within the Northumbria police force area. It uses the tool and has provided data for testing and development. We are interested in working with more local authorities and stakeholders in helping them make the best use of their traffic collision data and increase local authority/police collaborations to reduce traffic collisions throughout the UK and further afield.

Our current research is exploring seasonal effects on collision rates, and examining collisions at a finer scale. Improving predictions for the short-term future is the goal.

In future, it may be possible to make accurate real-time crash predictions, bringing traffic management and law enforcement closer to eliminating collisions that cause fatalities and serious injuries. Accurate collision prediction has an integral role to play in helping to make our motorways and cities truly smart.

Read the full article in PSE.

Time lapse of Tyne Bridge

published on: 30 October 2019