What does Waymo, formerly Google’s self-driving car project and now a standalone company, want to be when it grows up?
It doesn’t want to be an automaker—CEO John Krafcik said that during his speech at the Automobili-D conference running concurrently with the 2017 Detroit auto show. Industry analysts seem to concur. Rather, Krafcik said Waymo will be a technology licenser. But the development progress reported by the former Hyundai executive and Ford engineer could lead some to conclude that it's aiming at the full-systems autonomous driving space currently dominated by global Tier 1s Continental, Delphi and Valeo.
“We’re building our own hardware suite with a fully top-to-bottom, full-stack approach,” Krafcik told the audience. That approach—classic auto industry vertical integration—includes all vision sensors, radars and LiDAR. The company also is producing its related “AI compute” artificial-intelligence platform in-house.
The comprehensive technology suite has been fitted to 100 Chrysler Pacifica hybrids Waymo purchased from FCA, which Krafcik said can deliver SAE Levels 4 and 5 (semi- and fully autonomous) driving and will soon serve as development vehicles on Arizona and California roads.
“We’re bringing a self-driving technology platform to market,” said Krafcik. “To be serious about creating fully self-driving cars that can help millions, we have to oversee both the self-driving hardware and software.”
Waymo’s design/engineering progress includes lowering the cost of its in-house LiDARs by 90%--to about $7,500 each—compared with the LiDARs that Google purchased a few years ago for use in early autonomous-vehicle testing, he claimed. Those units cost approximately $75,000 each.
Waymo’s approach to date runs counter to the industry trend of partnering—OEMs with Tier 1s, Tier 1s with subtiers and with emerging start-ups—as well as outright company acquisitions. All have been necessary for companies to build the technology resources required for taking fully validated and robust autonomous driving systems to production.
“By controlling the entire ‘stack’ they can optimize the software to 'plug holes’ in their hardware performance while it’s evolving and improving—and to optimize the entire system,” observed Sam Abuelsamid, a technology analyst with Navigant Research. “I don’t think they’re looking to sell systems a la Conti and Delphi.”
A veteran OEM technology-strategy executive who requested anonymity agrees. “I’m not convinced that Waymo will be able to produce the tens (or hundreds) of thousands of units that the automakers will want to contract for at the desired prices,” the executive added. “Conti and the others have worked successfully in that realm for decades. Also it’s likely that automakers would go with a trusted supplier.
"Right now, Waymo’s only ‘customer’ is FCA,” the executive noted. “When it’s all done, Waymo will be a software supplier to the others.”
Waymo has run 2.5 million miles of public-roads SAE Levels 4 and 5 testing to date, and expects to reach 3 million miles by May, Krafcik said. System and component reliability and scalability are the top focus for engineers, he noted.