Nauto’s mission is to make driving safer and smarter. As we drive towards achieving zero collisions, we know that collisions still happen and can create a costly burden on fleets. From timely claims processing to managing unreported collisions, fleet managers need a reliable collision and claims management solution in order to keep vehicles on the road and maintain productivity, exonerate drivers in the event of wrongful claims, and help protect the safety of their drivers.
With over 250 million AI-processed video miles and counting, we’re constantly expanding our artificial intelligence (AI) model library to both detect new driving risks (such as tailgating) and improve existing detection capabilities— including collision detection. In this three-part series, we’ll share perspectives and insights from our Data Science, Artificial Intelligence, and Product teams on how we detect collisions and equip fleets around the world to act quickly to help keep their drivers safe and vehicles productive, starting with what we’ve learned about collisions on the road today.
Using G-Force to Detect Collisions
Traditionally, collisions are detected based on change in velocity, measured in units of gravitational force (g-force). Since high-speed collisions are typically well above 2 g, many fleet solutions use sensor-based devices to detect g-force as their sole signal for collision detection.
However, in the case of minor collisions, such as a low-speed rear-ending or a sideswipe, much of the collision force may be absorbed by the bumper or frame of the vehicle, and not transferred to sensor-based devices located inside the vehicle. These minor collisions often register g-forces less than 2 g, and in many cases less than 1 g.
The G-Force Distribution of Collisions
To help us better understand the prevalence of these minor, low g-force collisions, we mapped the g-force breakdown of all the collisions we observed in Nauto-equipped vehicles during 2018.
We found that 43% of collisions detected by Nauto were less than 2 g, and 16% of all detected collisions registered a g-force less than 1 g. In other words, nearly half of the collisions detected by Nauto were minor collisions with a g-force signal comparable to normal driving operations, such as turning into a parking lot, which registered 0.953 g in this example:
So how can we help fleets quickly identify when minor collisions actually occur? Stay tuned for our next installment on how we’re leveraging AI and machine learning to detect collisions across all g-force levels in real-time. Can’t wait? Contact us to learn more!