TU-TESL

• 22

According to foreign media reports, Israeli company nontraffic has developed a turnkey traffic management platform to support V2I (vehicle to infrastructure) communication and optimize signal control intersections in real time. The system is based on a plug and play artificial intelligence sensor intelligent network, which can transform each signal control intersection into an intelligent intersection. In less than an hour, users can manage traffic flow in real time. (photo source: nontraffic official website) this active method is in contrast to the current common intersection technologies such as inductive loop traffic detector, which is installed underground (hard to upgrade or replace) and can sense cars like metal detectors. Moreover, such traffic detectors are also limited by the fixed time schedule of intersections. Although the detector can minimize or maximize the duration of traffic lights, it cannot override them, even if no traffic is coming in that direction. Nontraffic's system deploys AI on the edge of the network, reducing installation and maintenance costs, and enabling intersections to prepare vehicles, not just respond to them. Nontraffic has AI sensors in all directions of the intersection. Yoav valinsky, a computer vision researcher at nontraffic, said the device uses NVIDIA's Jetson platform and GPU acceleration framework to integrate computer vision with radar, processing data about 15 times a second. These sensors also integrate network vehicle functions based on DSRC and c-v2x technology. DSRC is a wireless communication channel system between vehicle and infrastructure, and c-v2x technology can allow communication between vehicle, infrastructure and any related entities. Nontraffic's equipment can detect and classify all road users at the edge, including cars, buses, trucks, bicycles and pedestrians. Then, the processed data flows into the optimization engine, which is installed in the traffic light control cabinet, which has already been installed in most intersections. The data in the engine is used to optimize and manage traffic lights, both at individual intersections and throughout the city network. By placing the engine closer to the operating point, the edge system of nontraffic saves bandwidth, reduces latency and achieves faster computing speed. Nontraffic safely sends the data of each intersection to the cloud for further processing and optimization of urban traffic network. In addition, nontraffic displays information in the Urban Engineer's design dashboard for big data analysis, remote monitoring of intersections, and implementation of new traffic strategies. The real-time analysis ability of the system provides the function of collision prediction. Ilan ROZENBERG, director of business development at nontraffic, explained that the sensor device calculates the speed, acceleration and direction of the vehicle, so it can infer when the vehicles cannot see each other and when they may collide. Sensors can classify vehicles and give priority to specific road users. If a city wants to give priority to public transport or pedestrians at intersections around the school in the morning of a working day, the city engineer can enter the relevant traffic strategy on the dashboard, and the system of nontraffic will automatically change it. Nontraffic is a member of NVIDIA's inception incubator program. It is currently focused on the U.S. market and has conducted pilot projects in several cities and counties in the United States. Nontraffic reduces delays by 2700 hours per year, and reduces carbon emissions by 33 tons per intersection.