Nvidia, the tech company that provides Uber with its chips for the AV platform, has decided to suspend all AV testing until the results of the fatal car accident in Tempe Arizona is concluded. Nvidia is involved in testing Self-Driving pilots around the world, and their chips can be found in cars in Germany, Japan, and the US.
A Nvidia spokesperson told the media that "The accident was tragic. It's a reminder of how difficult SDC technology is and that it needs to be approached with extreme caution and the best safety technologies. This tragedy is exactly why we've committed ourselves to perfecting this life-saving technology. Ultimately AVs will be far safer than human drivers, so this important work needs to continue. We are temporarily suspending the testing of our self-driving cars on public roads to learn from the Uber incident. Our global fleet of manually driven data collection vehicles continues to operate."
Toyota, the Japanese car giants, also reported their halting AV testing in lieu of the psychological and emotional stress the accident has had on their safety drivers. In fact, one question has not been asked, what are the procedures that a safety driver must perform while sitting in the car? This is not so important to the Nvidia issue, but the answer is complex and requires the definition of the role. Perhaps this is what Toyota is really doing, sopping the testing to evaluate what the safety driver must do while seated in the car.
Back to Nvidia. Nvidia is a major tech company that provides solutions to computing platforms, graphics interfaces, and AV technology. They developed the Xavier chip specifically for the AV market, and according to NVidia, have 320 clients, which roughly translates as everyone involved in AV development.
The three major chip manufacturers involved in the AV market are Nvidia, Intel, and Arm. Intel owns Mobileye, Nvidia is in collaboration with Uber, and Arm is a SoftBank company, so you know where its technology will go. The bottom line is that the chips involved are the core of any successful processing solution. This means that all the LiDAR's radars, cameras and what have you's are all sending their data to a central processing unit (CPU) which is a lot of microchips being operated by software. The integrity of the chip is as integral to the success of the operation as the software management system coded onto the hardware. Nvidia would prefer to know for a fact that their chip was not the fault of the accident, and as such will wait to find out of the software did not make a decision in the time it was expected to make that decision. Or, was there a delay in the transfer of information from the sensor arrays through the hardware (Nvidia chips). The ultimate question of who is to blame, software or hardware?