Accident statistics are a sure-fire way of measuring the safety of a road. The numbers don’t lie, after all. However, there is one essential set of statistics that, until now, have gone largely unrecorded: near misses. Near misses in the world of aviation are investigated thoroughly, so why aren’t they taken into consideration on our roads? Intertraffic caught up with two companies working together using AI to ensure that this is no longer the case.
“Crash data is reactive and often does not contain enough serious risk information for reliable diagnostics and effective improvements. The serious crashes occur in extremely small clusters, as a result, it is hard to diagnose risk factors from historical data,” says Winnipeg, Canada-based MicroTraffic CEO and co-founder Craig Milligan. “What we do is combine computer vision with road safety and injury modelling to unlock the precise, predictive, and proactive diagnostics of serious latent risk factors at intersections. Transportation engineers can turn these precision diagnostics into targeted and effective road safety investments.”
The technology measures five variables for every risk event: speed, angle, generalized temporal separation, user vulnerability, order of interaction. Near-misses at intersections are used to proactively identify opportunities for safety improvement by applying AI analytics to video feeds from existing traffic cameras.
“Based on the detected risks, we help cities visualize implementation of reflective materials on safety countermeasures for all road users,” says Milligan.
3M and MicroTraffic may seem, on the face of it, unlikely bedfellows, but a chance meeting with 3M Transportation Safety Division President Dan Chen at a conference and the subsequent discovery that the companies’ complementary approach to saving lives and that MicroTraffic could help 3M on their noble quest.
Fast forward through the pandemic and Milligan and Sauter presented a paper at Intertraffic on Wednesday (AI-Driven Intersection Safety Diagnostics & Proven Countermeasures Using Retroreflective Materials) that focused on their work to date. Perhaps the companies’ most notable collaboration to date has seen them partner with Austin Department of Transportation in Texas to automatically measure near-misses at several high-risk intersections using traffic video. The system measures near-misses for all modes of transportation and, says Austin DoT, promotes proactive understanding of intersection risk.
“What we’re talking about is a silent risk,” says Milligan. “And you need to fix that risk while it’s silent – you want to fix that dangerous intersection before the first fatality occurs, not after it. Our solution uses computer vision to measure the road users’ movements and then kinetic energy risk models to assign near miss risk scores to those movements.”
The question of how all this incredibly valuable data is turned into insight and information is Gernot Sauter’s territory. The head of 3M Transportation Safety Division’s Senior Regulatory Affairs Specialist near Dusseldorf explains that his goal in projects such as this is to gain information on the effectiveness of 3M’s retroreflective safety devices.
“The evidence data is coming from MicroTraffic – by analysing the near misses you get some statistic evidence about what intervention works, what is most successful and also what has perhaps not worked so well. This is where we come into the picture – we can use the data provided by Craig and his team.”
For MicroTraffic’s distributor in Japan, Sam Miyagawa of Miyagawa Consulting, the task was straightforward: find an accurate, reliable method to record and analyse traffic near misses. A visit to a traffic safety conference in Louisiana was clearly the answer.
“Japan is a relatively small country but we have a lot of traffic accidents. We came across MicroTraffic at the ATSA event in New Orleans a couple of years ago and we quickly realised that we could certainly use their technology in Japan. We work with about six major consulting companies – they send video and we forward it on to MicroTraffic for analysis.”
Milligan concludes with his assertion that for road authorities the only way forward is to be proactive and not reactive.
“We’re finding that more and more transportation departments are wanting to move to a proactive, preventative approach. They don’t want to wait for more fatalities before taking action – and that can only be a good thing.”
Sam Miyagawa (Miyagawa Consulting), Craig Milligan (MicroTraffic) and Gernot Sauter (3M)