NVIDIA has announced that they will install their revolutionary self-driving system, the Drive PX 2, in all cars participating in the Roborace Formula E Series.
Roborace Formula E is an all-robotic electric car variant of the Formula One slated to begin early next year. The announcement was made at the GPU Technology Conference in San Jose, California. Jen-Hsun Huang, CEO of NVIDIA stated that the company’s Drive PX 2 system would be standard in all cars managed by the Roborace teams.
The Drive PX 2 can be held in one hand and has the capacity to perform 24 trillion operations per second and extract data from radar sets, LIDAR and a dozen cameras. Michael Houston, technical lead, stated in an interview to IEEE Spectrum that the system could glean knowledge from deep learning – a reiterative analysis method that can discover patterns in mass quantities of data.
“Deep learning has different applications,” said Houston. “The focus has been on the visual analysis of imaging in video—web science, embedded systems and automotive. Fundamentally, we’re processing pixels.”
Roborace will increase the intensity and excitement of a car race by including autonomous cars. The absence of a driver allows for these cars to be made sleeker, faster and much more maneuverable. Designers could even install wind-tunnel apertures in place of the cockpit. The cars will be identical in appearance although the software in them will vary significantly.
AI will play the most important role in the race. Theoretically a calculating algorithm that runs on silicon and without the encumbrance of carbon-based system should be able to pilot a car in a much more intense manner pushing the boundaries of speed and maneuverability. However, a test drive conducted last year saw Spectrum’s Lawrence Ulrich beating Audi’s robotic RS7 race car on a track near Barcelona by 4 seconds.
Roborace is an interesting platform for companies to showcase the capabilities of autonomous cars. The contest will be a great opportunity to unearth new ideas and techniques of enhancing the automobile industry further.
Illustration by Daniel Simon/Roborace/IEEE Spectrum