While the discourse on Artificial Intelligence (AI) continues, one Stellenbosch company is solving real world problems with its technology in a sector that affect all locals – managing traffic.
Bytefuse is an AI research and development company that forms part of the Alphawave Group, based in Technopark. What started as research for Syntell, a provider of road-safety and fine-management services, has turned into a traffic solution that operates in real time. Quebit is a smart traffic system designed to enhance traffic infrastructure within cities.
Quebit uses AI to read the traffic flow and respond accordingly by coordinating traffic lights. “Making every single traffic light on the R44 green also isn’t the answer,” warned Greg Newman, Bytefuse CEO.
According to Newman Quebit’s aim is to manage traffic flow to relieve congestion, creating a more efficient commute for anyone behind the wheel. Configuring the timing of traffic lights has been based by collecting data in person. “It would have a counter and take a snapshot of time where it counts the vehicles and then uploads that data into some sort of rough algorithm, and it spits timings out,” Newman explained.
He said in using this process the chance of human error is high, yet the technology has been used for years – and it does not take changes to a town’s traffic patterns into account. Changing such patterns and adapting traffic systems can be slow and expensive.
Newman said: “We were wondering if it wasn’t possible to use AI to start reconfiguring these intersections automatically, using a system that functions optimally? So we spent quite a while working in simulation.”
After a while a Quebit pilot project was launched in Stellenbosch, where congestion and traffic delays has become synonymous with the town.
According to Shahil Mawjee, head of engineering at Bytefuse, the focus was on gathering data for the route between Stellenbosch and Somerset West along the R44. The first intersection fitted with the QueBit was at the entrance of Technopark.
The true test of the AI technology came when traffic was at its peak. In this case, it was at the time of an event held at Blaauwklippen, its main entrance at the Technopark intersection, which caused back-up traffic at the wine estate. “There was an enormous number of cars going in and out of Blaauwklippen, and the timings of that intersection just weren’t equipped enough to deal with that kind of volume,” Mawjee pointed out.
According to the Bytefuse team such real-time observation enables them to make corresponding updates to a technology that becomes indispensable. “We found that whenever we took control, we were able to clear those queues in two traffic cycles.”
This process proliferated; data was gathered from five intersections in all, with QueBit added to various intersections, including Blaauwklippen/Technopark and Annandale. Synchronising these intersections opened up new possibilities and the team found travel times between Stellenbosch and Somerset were significantly cut down, from 14 minutes to three.
Traffic volumes during peak times and short-term traffic patterns were all captured during the three-week trial. Among the finds was that road users on the monitored route held to the same travel times from Monday to Thursday, but on Fridays the exodus out of Stellenbosch, tended to start much earlier, at 14:00.
According to Cobus Louw, Bytefuse machine learning engineer, the data gathered through Quebit can assist municipalities in the short and long term. It informs future planning and long-term infrastructure projects using an updated data-driven approach.
He added that possible uses for Quebit are vast – addressing safety concerns, learning road-user behaviours, all thanks to AI.
The Quebit team is now expanding their software and gearing-up to role out their AI solution to the market.
* Bytefuse is part of Novus Holdings. Eikestadnuus is part of Novus Media, a subsidiary of Novus Holdings.
