New technology installed on traffic signals could soon help prevent collisions at intersections.
Derq, a transportation startup in Dubai, United Arab Emirates, has melded machine-learning algorithms with information from cameras and radar sensors attached to traffic infrastructure to detect red-light runners and other imminent hazards.
Derq's software monitors vehicles, pedestrians, bicyclists and other road users. Within two seconds after a potential conflict is detected, signals sent to properly equipped vehicles could issue warnings to motorists.
Derq expects to feed analytics to transportation departments and other partners so they can learn more about trouble spots.
Data on near misses can be a more useful measure of safety than information on collisions.
"At best, crashes are a small sample size and reactionary," said Anthony DeFilippo, director of business development at Derq. "We're going to something more predictive and informative of longer-term trends."
City and regional leaders believe data analytics can help reduce traffic deaths and injuries. With funding from PlanetM, the mobility division of Michigan Economic Development Corp., Derq has run a pilot project for the past year at Jefferson Avenue and Randolph Street in downtown Detroit. In August, the startup announced plans for a similar project in Dublin, Ohio, where it will work with Denso. Elsewhere, Derq has partnered with Qualcomm.
Starting in Dublin, Derq plans to gather data from intersections and stitch it together to make safety inferences along corridors. For example, if a driver runs a red light, the technology may be able to alert drivers farther down the road of a potential hazard.
"That's the power of using a machine-learning approach as opposed to a rules-based one," DeFilippo said. "We're better able to understand if they're likely to run another one."
Derq's technology is used at 10 intersections in the U.S., Dubai and Austria. DeFilippo says the company plans to boost that to more than 30 by year end.