Nvidia has another design win for its Drive PX artificial intelligence platform for autonomous driving: Toyota. Nvidia CEO Jensen Huang made the announcement this week in the opening keynote of the company’s GPU Technology Conference in San Jose.
Coming on the heels of Nvidia hookups with Mercedes-Benz and Audi, it’s a clear sign that the once-just-a-graphics-card company is serious competition in the automotive sector for the likes of Intel, which plans to invest $15 billion on autonomous driving.
Helping self-driving cars get smarter
The Nvidia system would mean Toyota could have self-driving cars available for sale “in the next few years,” Huang said Wednesday. That jibes with what some automakers–those working with NVidia and those aligned elsewhere–have said. 2020 is the earliest date bandied about. Others in the industry have said privately that serious self-drive cars may need another 5-10 years for ones that would never ever need someone behind the wheel just in case.
Nvidia first announced its Drive PX platform in 2015. Drive PX 2 was announced a year later at CES 2016. Nvidia describes Drive PX as an open AI car computing platform that enables automakers and their tier 1 suppliers to accelerate production of autonomous vehicles: “It scales from a palm-sized, energy efficient module for AutoCruise capabilities, to a powerful AI supercomputer capable of autonomous driving.” A simpler, single-processor version allows for “AutoCruise,” or automated highway driving and HD mapping. Multiple platforms in parallel would allow for fully autonomous driving, meaning self-driving on any road.
Previously, Mercedes-Benz and Audi agreed to work with Nvidia on autonomous driving. Now comes Toyota, the world’s second-largest automaker, with 10.2 million vehicles produced last year, just behind Volkswagen (which includes Audi) with 10.3 million vehicles produced.
Nvidia AI lets cars make own decisions
At CES 2017 in Las Vegas, Audi showcased an Nvidia-equipped, self-driving car that made unassisted loops around a nearby parking. Nothing tricky there, until an Audi tech ran onto the track and stuck a big Road Closed sign with arrows pointing left. Encountering the changed road conditions, the car responding by taking a left turn and diverting itself through a set of cones and then back on-course.
Virtually every automaker is working on autonomous cars. All are doing significant RD work on their own, but none appear ready to create all the optical, radar, and lidar hardware from scratch, let alone the special-purpose CPUs and algorithms. It is not unusual in the auto industry to work with third parties to provide significant parts of the car, such as transmissions, infotainment head units, lane departure warning, blind spot detection, and adaptive cruise control.
Nvidia’s competition includes Intel and Qualcomm. There are also others creating or pondering autonomous driving technology, including Apple, Google, and Amazon. It’s possible some of their work could be licensed to automakers. Where Tesla built its own cars from scratch, Apple in the past has looked for a partner to manufacture its autonomous car if it goes ahead.