A key executive in the autonomous driving space said he gives Waymo the lead over Tesla in the self-driving software journey.
Ali Kani, vice president of automotive for chip maker NVIDIA, which is considered the world’s most valuable company, told a group of automotive journalists that Waymo’s redundant systems and diversity of sensors are key to that company’s success so far.
“If any single thing goes wrong on the car, there’s a backup,” Kani said Wednesday, Oct. 7, during an Automotive Press Association webinar with David Welch of Bloomberg. “Even if they can’t complete the mission with the car, there’s a backup car.”
Waymo, incidentally, announced this week that its
operations in Detroit would now include fully autonomous drives without a safety driver at the wheel. The company said it plans to launch public ride-hailing in the future, but for now, rides will be limited to Waymo employees.
Waymo already offers ride-hailing in a number of cities, including Phoenix, San Francisco and Los Angeles. Its operations haven’t been without incident, but Tesla’s robotaxi rollout has struggled by comparison, with the recent news that its vehicles have trouble seeing pets at night.
Tesla CEO Elon Musk has long promoted the notion that lidar isn’t necessary for self-driving vehicles and has focused on the use of cameras to detect objects instead. Many self-driving industry experts, however, say redundancy that incorporates cameras, radar and lidar is important for safe operation. Lidar, usually seen as a device spinning atop a self-driving vehicle, bounces lasers off an object to create an image.
Tesla is “world-class” in camera detection in a way that can make one reassess what’s possible, but, Kani said, radar and lidar are good at detecting objects regardless of issues like weather. It’s a key point in the race to capitalize on the self-driving space. NVIDIA, he said, sees a “$10 trillion” opportunity in the self-driving space, providing its technology to companies across the spectrum.
That includes Tesla and Waymo but also ride-hailing companies such as Uber and Lyft as well as automakers Stellantis, Nissan, Mercedes-Benz, Hyundai, Nissan and Volkswagen. NVIDIA has been working on self-driving technology since 2015 and supplies training and simulation technology as well as the in-car “self-driving stack,” although not all companies opt for all three.
One of Kani’s slides during the presentation said the “robotaxi industry runs on NVIDIA." Another slide listed NVIDIA’s global robotaxi rollout as 28 locations across eight countries in North America, Europe, Asia and Australia by 2028.
Some have questioned the viability of self-driving vehicles and robotaxis as a business model, a topic Kani was asked about during the discussion. General Motors, for instance, ended up folding its Cruise subsidiary into its engineering division, and Ford Motor Co. and Volkswagen shuttered Argo AI.
Self-driving cars used to be custom built, with sensors costing tens of thousands of dollars, but lidars, for example, have dropped in price, costing hundreds of dollars now rather than thousands of dollars, he said.
The cars themselves might cost $50,000 today, meaning a robotaxi company would need to account for that initial investment as well as maintenance and fleet management. But that would still leave money to be made.
A self-driving car could travel 75,000 miles a year at an average price of $4 per mile, bringing in about $300,000 per year, he said.
“It’s a healthy business model as the economics scale,” he said.
Eric D. Lawrence is the senior car culture reporter at the Detroit Free Press, and he also writes about recreational vehicles, boats and bikes. Send your tips and suggestions to elawrence@freepress.com. Become a subscriber. Submit a letter to the editor at freep.com/letters.
This article originally appeared on Detroit Free Press: NVIDIA exec says one company has the edge on self-driving cars
