Bringing together Transportation, Statistics and Software experts to reduce injuries and fatal accidents on the roads worldwide – that is the idea of our Statistics and Road Safety research group at Newcastle University. Jointly with PTV we are developing cutting-edge statistical methods, and user-friendly software applications, for road safety practitioners to analyse road collision data and to predict traffic hotspots. In addition to that, we educate and train the next generations of road safety engineers through our postgraduate taught and research programmes.
The goal: Predicting and preventing accidents
The first objective of our research was to develop improved ways of predicting where, on the network, road collisions are most likely to happen. This will enable road safety practitioners to be proactive, rather than reactive, in terms of collision prevention. The second objective was to provide improved methods for more accurately evaluating the effectiveness of road safety interventions to guide future investment decisions in road safety. Finally, we were very keen to take these techniques ‘out of the laboratory’ and make them available to road safety practitioners worldwide through easy-to-use software applications.
“Reducing the number of people being killed or injured on our roads – that’s the goal of our Statistics and Road Safety research group at Newcastle University.”
Conducting a before-and-after-study
The research began by using collision data to evaluate the performance of mobile safety cameras in reducing the number and severity of collisions in a before-and-after study. This project used a Fully Bayesian framework for taking account of confounding factors that can often exist in road collision data. These factors make it very difficult to separate out the ‘true’ scheme effects from the non-scheme effects. We developed algorithms to implement this approach and test it with our road collision data.
Turning a research project into a software solution
At this point we began discussing our research with colleagues at PTV and, over time, together we adapted the before-and-after approach to be able to begin predicting where collisions hotspots might occur in the future. New algorithms were developed that became embedded in PTV’s VISUM Safety software package to satisfy our objective of making the research outputs usable by road safety practitioners. We have also continued to develop an internal version of the software as a research and development tool to test out new and emerging ideas in hotspot prediction and scheme evaluation.
“It is great to see that our research has a real-world influence on reducing deaths.”
We are very pleased that the results from the validation exercises we have undertaken so far, in a number of cities around the world, suggest that our algorithms are able to predict future collision hotspots with a very encouraging degree of accuracy compared to other approaches. Furthermore, road safety practitioners have found this information extremely valuable in their decision-making processes regarding where to target road safety investment on their networks. It is great to see that the outputs from our research have a real-world influence on reducing death and injuries on our roads.