On January 11, 2021, Transportation Research Record, the prestigious peer reviewed journal of the Transportation Research Board of the U.S. National Academies of Sciences, published findings by Dr. Bhagwant Persaud at Ryerson’s University who studied the ability of MicroTraffic’s ability to predict injury crashes using safe systems surrogates.
The compelling results showed a predictive accuracy rarely seen in attempts to model crash counts.
The key to MicroTraffic’s ability to proactively detect intersections and specific movements at risk of crashes is the kinetic energy approach to measuring near-misses.
For a long time, near-misses were defined by the temporal separation between road users. Researchers debated the correct measure of temporal separation (post encroachment time, time to collision, T2, etc) and they debated the correct threshold for a severe conflict.
The problem with the focus on temporal separation is that it only sheds light on the likelihood of a crash, not its severity. Crash severity depends on forces applied to a human body, from the dissipation of kinetic energy, which exceed the biomechanical injury thresholds of a person.
When MicroTraffic was founded in late 2017, we focused on a kinetic-energy based model from the start. We used research on injury biomechanics, impact angles, and user vulnerability to construct risk indicators that captured both the likelihood and the severity of a crash.
Now, with 65 communities using MicroTraffic technology, this validation from TRB and Ryerson University researchers is a tremendous encouragement to us and the cities relying on our data. When we flag serious near-misses, it means that a serious injury is imminent and the road authority is justified in making an investment to proactively prevent injury from happening. When we followup and measure a reduction in serious near-misses, the road authority can know the safety impact that they achieved.
Also of encouragement to us is that when cities follow our intersection safety plans developed with video diagnostics, they measure on average and 80% reduction in near-misses. This is meaningful because those near-misses are strongly correlated to crashes.
The paper is published here for those with institutional access.
For an unofficial version, please contact sachin@microtraffic.com with your request.
