Artificial Intelligence may have the answer to our traffic woes, according to an in-depth study supported by Nvidia. Every morning and most evenings, scores of commuters are trapped in gridlock traffic. The reason? An increasing number of cars are on our roads, and the infrastructure cannot accommodate them. So what's the solution?
Rogers Communications, a Canadian telecommunications company, has been looking at solving congestion using AI for some time. The company recently deployed a 5G-connected, AI-enabled smart traffic system at the University of British Columbia (UBC) and managed to reduce traffic and bring other improvements. NoTraffic collaborated with Rogers Communications on the project.
The computing system was able to calculate "tens of thousands of possible traffic flow scenarios" for each road user and select the best possible one in real time. This simultaneously decreased traffic and emissions.
Importantly, safety at intersections saw an improvement. The AI could automatically prioritize first responders, public transport, and vulnerable road users such as cyclists. Traffic light operations can also be delayed or extended to reduce motorist irritation and keep traffic flowing smoothly.
Additionally, the AI can glean useful analytics that enlighten researchers about things such as 'arrival on green' times, side-street efficiency, and even traffic equality. Furthermore, the system can provide real-time reporting on dangerous behavior, such as speeding and running red lights.
Here's how it works. Edge sensors, powered by Nvidia's cutting-edge technology, could detect traffic and differentiate pedestrians and cyclists from larger vehicles. Based on the live data, the AI could adjust the traffic light timings in real-time, thereby improving traffic flow and safety.
The experiment was conducted at five junctions on the UBC Campus with heavy foot traffic.
"Using the Nvidia Jetson platform, we can provide a highly accurate and reliable neural network-based solution that detects road users at a super-human level in all weather conditions," said NoTraffic's Chief Technology Officer Uriel Katz.
The results of the experiment cannot be ignored. Over one year, wait times for pedestrians were reduced by 2,500 days, while motorist wait times decreased by a staggering 4,700 days. The environment benefits, too, as carbon dioxide emissions are estimated to have declined by 75 tons, with "an estimated economic advantage of $165,000."
One can imagine the benefits if this was employed on a broader scale. And this could happen soon. Rogers and NoTraffic are working together to make this AI technology available to stakeholders who will benefit from it.
While reduced wait times and less traffic are lovely, the improved safety of this system is worth its weight in gold. America's road death figures are still ridiculously high, so a system like this can potentially reduce vehicle collisions and save additional lives - especially vulnerable pedestrians at intersections.
This is a good idea, and we wonder how it would work with the advanced systems car manufacturers are working on. General Motors, for example, is working on a Car-to-X system that could communicate with traffic lights. The idea would be that your Cadillac Lyriq or ultra-luxury Celestiq would never be stuck at the light again.
The system works with GM's sophisticated Ultra Cruise system and can slow or speed the car up depending on the traffic lights ahead. This would reduce fuel consumption, save time, and relax the driver. Clearly, this is an avenue of artificial intelligence worth pursuing.
