Skip to content

Metropolis Spotlight: Bluecity Combines Vision AI and Lidar for Real-Time Road Safety and Traffic Congestion

Cities now have access to real-time, multimodal traffic data that improves road safety and reduces traffic congestion.

Bluecity, an NVIDIA Metropolis partner, recently launched a new traffic management solution for safer roads and shorter commutes. The technology combines vision AI and lidar technology to better understand round-the-clock traffic data, providing information that could help city planning departments identify problem intersections, reduce congestion, plan smarter, and lower emissions. 

Road safety and congestion are a priority for city and transportation planners, but sparse data has limited their ability to address traffic issues, especially in areas with ever-expanding populations. While video cameras are used to capture information, like the number of cars at a particular intersection, poor lighting and bad weather conditions can interfere with capturing this data accurately. Studies have also shown that accidents are most likely to happen during those times when visibility is low.

New technologies that can overcome these obstacles and collect multimodal data about drivers, vehicle speed, and trajectories can help make the roads safer. Especially when it can be done in real time. 

Bluecity is solving this problem by combining vision AI and lidar technology to understand and evaluate traffic data. 

Figure 1. Bluecity analytics platform

Similar to radar, lidar sensors emit pulsed light waves into the environment and sense objects when the pulses bounce off them. Lidar uses lasers with a lower wavelength than radar, and as a result can detect smaller objects, offering precise measurement data. Lidar can do this even in poor lighting or weather conditions and it captures data anonymously. Each lidar sensor can provide 360-degree coverage and a radius of up to 120 meters (400 feet).

The Bluecity system employs the powerful capabilities of the NVIDIA Jetson edge AI platform—which provides GPU-accelerated computing in a compact and energy-efficient module—along with NVIDIA TensorRT to accelerate its application’s inference throughput at the edge. The edge computing system runs a proprietary 3D perception software and powers the traffic management solution to process up to 50 lidar frames per second in real time to detect all road users.

Their platform provides information on which turning directions and intersections are the riskiest, near misses, time-to-collision, and the speed of the vehicles involved.  It can also classify road users and gives important insight into not only driver behavior, but also that of cyclists and pedestrians. 

The ability to collect data regardless of lighting or weather conditions helps city planners make decisions about things like road design and traffic-light timing that are based on actual data. The startup’s AI component turns raw data into valuable information that guides practical and timely decision-making. AI powers better visualization, while conflict analyses help identify dangerous intersections before accidents occur. For instance, in Repentigny, Quebec, Bluecity installed their solution to provide multimodal traffic data so an engineering firm could better understand mobility in a region where they are updating a bridge. 

Subscribers can view, select, filter, and download the data in order to improve their traffic planning with Bluecity’s easy-to-understand dashboards.

Bluecity solutions are deployed in several Canadian, U.S. and European cities, including Irvine, Austin, Texas, Boca Raton, Trois-Rivières in Canada, and Helsinki, Finland. The startup’s vision is to provide better multimodal data to make intersections in our cities safer, improve road safety, and lead to reduced carbon emissions for smart cities.

Source:: NVIDIA