At the 2019 GPU Technology Conference, Nvidia CEO Jensen Huang announced that Toyota’s Japan-based research arm is using Nvidia’s full end-to-end development and production to develop, train and validate the automaker’s autonomous vehicle technology.
Toyota is already using the Nvidia Drive AGX Xavier computer to power self-driving vehicle testing in the real world. Nvidia collaborated with Toyota Research Institute-Advanced Development (TRI-AD) in Japan as well as Toyota Research Institute in the United States. This collaboration makes Toyota the first customer to try out Nvidia’s just-announced “Drive Constellation” platform. The new cloud-based, autonomous vehicle simulation platform will let self-driving car developers run tests on virtual, rather than real roads.
With the new platform, Nvidia said, autonomous vehicles can “operate in situations ranging from routine to dangerous, with greater efficiency, cost-effectiveness, and safety than what is possible to achieve in the real world.” The system consists of two distinct cloud-powered solutions: "Constellation Simulator" server and "Constellation Vehicle" server. The Simulator, with the help of NVIDIA's GPUs, will run software called ‘Drive Sim’ to generate a realistic virtual world that would be fed into the virtual car’s sensor. On the other side, the Vehicle server – powered by the Drive AGX in-car computer – processes the simulated sensor data. The driving decisions thus derived from the vehicle are fed back into the Simulator “to realize bit-accurate, timing-accurate hardware-in-the-loop testing," said NVIDIA.
“Our vision is to enable self-driving vehicles with the ultimate goal of reducing fatalities to zero, enabling smoother transportation, and providing mobility for all,” TRI-AD CEO James Kuffner said in a statement. “Our technology collaboration with Nvidia is important to realize this vision. We believe large-scale simulation tools for software validation and testing is critical for automated driving systems.”