NVIDIA's founder and chief executive officer Jensen Huang recently clarified what type of technology the company provides to Uber's autonomous driving program and asserted that the firm's self-driving platform was not used in the recent fatal incident that saw a driverless Uber kill a pedestrian in Tempe, Arizona. The NVIDIA chief told CNBC that Uber's self-driving technology is different than that of NVIDIA, saying the former developed its own sensing and driving software.
Huang's statement comes two weeks after a self-driving Uber car hit and killed a woman as part of the world's first autonomous vehicle accident that resulted in a pedestrian death. The woman was reportedly crossing a road outside of a crosswalk and the self-driving vehicle was unable to detect it for unknown reasons. The accident occurred despite the fact that a human backup driver was in the autonomous vehicle, as required by local regulations, though they failed to take over the car prior to the fatal accident. The San Francisco-based startup immediately suspended all of its autonomous driving tests in the country following the incident. NVIDIA is one of the companies that recently teamed up with Uber for the ride-hailing company's self-driving push. At CES 2018, the American technology company confirmed that it had been selected to power Uber's fleet of self-driving vehicles. As part of the partnership, Uber will use NVIDIA's chips to run AI algorithms and ensure it has access to the computing power needed to guarantee the safety of its passengers.
While clarifying that Uber was not using NVIDIA's self-driving technology per se, Huang also reiterated that the company upholds its decision to suspend all of its autonomous driving tests until Uber sorts out a way to solve its autonomous driving dilemma. Huang also said NVIDIA will try to gain insights on the investigation and learn how it can avoid the same problem with its own self-driving technology in the future. Two years back, NVIDIA unveiled its proprietary supercomputer system for self-driving cars called the NVIDIA Drive PX 2. It can achieve eight teraflops of computational power driven by twelve cores built on a 16nm FinFET process and one of NVIDIA's super-powered Pascal GPUs based on the Maxwell architecture.