NVIDIA's Jetson Nano Takes Pole Position at the DIY Robocars Race
NVIDIA boasts of major wins in the robo-racer community, with the Jetson Nano largely supplanting the Raspberry Pi.
NVIDIA has been showcasing the prowess of its Jetson family of edge artificial intelligence (AI) devices by competing in the DIY Robocars community — even as our own AI at the Edge Challenge, powered by the NVIDIA Jetson Nano, continues to attract applicants.
Robo-racing, in which autonomous scale vehicles attempt to outdo each other on the track, is a growing sport — some 10,000 participants across 40 countries, 3DR chief executive and DIY Robocars host Chris Anderson estimates. There's been a sea-change of late, though. The use of NVIDIA's low-cost high-performance Jetson Nano computer-on-module (COM). "Jetson Nano is definitely starting to be adopted as a maker platform," claims NVIDIA's John Welsh, who participated — and placed — in a recent robo-race in California using just that device as a driving system. "It’s an example of this paradigm shift to AI and how this maker community is taking the leap."
"On a recent Sunday in September, a couple hundred enthusiasts gathered at the Circuit Launch space in Oakland, Calif., for a robo race," writes NVIDIA's Scott Martin. "Just months earlier, many in this maker community of do-it-yourself racers were using Raspberry Pi, but now they’ve mostly switched over to Jetson Nano to stay competitive. NVIDIA’s team took first place in the 'stock' car race. In the 'unlimited' race, which allows bigger-budget cars, Jetson was used in the cars that placed first and second."
Martin points to the high performance of the Jetson Nano, which delivers up to 472 GFLOPS of compute in a 5W envelope, and its support for popular AI frameworks including PyTorch, Keras, TensorFlow, and OpenCV along with NVIDIA's own CUDA-X library collection, as being key to its success in the field.
The DIY Robocars event wasn't NVIDIA's first time strapping wheels to a Jetson Nano, however: The company launched the COM, which is usually purchased with a motherboard to turn it into a standalone development system with USB, networking, and video outputs, alongside an open source autonomous robot design dubbed JetBot - and has since released the JetRacer project to get would-be robo-racers up and running as quickly as possible.
Martin's full write-up is available on the NVIDIA blog, while more information on DIY Robocars can be found on the official website.