GM bought a little-known startup and suddenly became a dominant force in self-driving cars
- General Motors and Cruise Automation made news by buying Pasadena-based Strobe, Inc.
- Strobe's CEO explained how the deal happened.
- She also provided details on the company's technology.
Julie Schoenfeld has seen the future of the self-driving, and it depends a single critical piece of technology.
"Everybody is kicking the tires on every Lidar company," the CEO of Strobe, Inc. said.
Lidar is "laser radar," and anyone who has seen what Schoenfeld, who started Strobe in 2014 and sold it last week to General Motors for an undisclosed amount, entertainers referred to a "spinning Kentucky Fried Chicken" bucket atop an experimental autonomous vehicle from Waymo or Uber knows what Lidar looks like.
What it does is another matter, but the technology is according to many in the incipient self-driving world critical to vehicles that will someday achieve full autonomy and be able to drive themselves with no human input,
Schoenfeld counts herself as a Lidar acolyte. "Lidar is absolutely required - and required right now," she said.
Strobe is just the latest very youthful startup to catch the attention of a major automaker bent of surviving the potentially epic business disruption that autonomous vehicles represent. The Pasadena-based company is slightly off-the-grid, however, in that it's located in Southern California rather than the Bay Area and has a connection with The California Institute of Technology rather than Stanford.
Sweet home Pasadena
But that's the way Schoenfeld, who previously served as CEO of Perfect Market before selling it to Taboola in 2014, likes it. She'll continue as CEO of Strobe and her entire team will go to work for GM, remaining in Pasadena. But there's a wrinkle: Strobe was effectively acquired not by GM but by Cruise Automation, the self-driving startup that GM bought for over a half-billion dollars in 2016.
Schoenfeld's new boss will be Kyle Vogt, Cruise's CEO. Cruise is based in San Francisco and is currently testing its integrated technology on a fleet of all-electric Chevy Bolts, with the goal of a roll-out in the Bay Area, most likely in connection with GM Lyft partnership.
The Strobe deal (which has already closed, with terms to be revealed when GM presents fourth-quarter earnings in early 2018) shows how much freedom Cruise has been given to operate within the sprawling GM structure.
"They all substance," Schoenfeld said of Cruise. "Very much like my team."
She shared a story about a meeting between Cruise and Strobe, with the telling detail being how much everybody wanted the meeting to end, constantly checking their watches. Not because they couldn't get along, but because they wanted to get back to work.
The anecdote matches up with Vogt reputation, both at GM and as an entrepreneur who has spent several years completely committed to making self-driving cars a reality. Vogt's goal is to work on solving one big problem, and to do so to the exclusion of everything else.
The Lidar problem
For autonomous vehicles, the problem is taking the driver out of the equation. A variety of semi-self-driving technologies have already entered the market - Tesla Autopilot and Cadillac Super Cruise are the two best - and Google's Waymo has amassed millions of self-driving miles, with most mishaps being mere fender-benders.
Uber dramatically unveiled a small fleet of self-driving cars in Pittsburgh last year, but the cars were distinguished be massive and probably quite expensive autonomous-technology rigs on their roofs.
So for the auto industry, the problem is actually twofold: eliminate the driver; and make sure that cars still look good.
Cruise has advanced the solution by developing hardware and software that can be built into a car during the assembly process, enabling a GM to use the technologies across its lineup of vehicles, just like an engine, a transmission, or an infotainment system.
The next stage of the process is where Strobe comes in.
"Lidar hadn't had the cost reduction needed," Schoenfeld said. But reducing cost carries with it a challenge: retain accuracy. Pointing out that basic Lidar technology has been around since the 1960s, she noted that if you can make a Lidar large enough and spend enough money, you can create a self-driving car. Unfortunately, that Lidar would be huge and cost a fortune.
But if you can make a Lidar small enough - and Strobe is bringing it down to microchip scale - you could lower the cost dramatically and have more Lidars per vehicle. Schoenfeld spoke of numbers like $100 per Lidar, with five units on each car. And Strobe's Lidars are already small; when the deal was announced, Vogt wrote a Medium post about it and included a photo of a Strobe unit next to a Sharpie pen for scale.
The Lidar advantage
The advantage that Lidars have over other technologies - such as the camera-driven system that makes up Tesla Autopilot - is that that they can create a detailed three-dimensional map of a vehicle's operating environment. As long as weather conditions are good, they can "see" far better than a human. Camera systems can, too, but they require huge amounts of computing power to process all the visual data. (Camera systems can also operate better in bad weather.)
But Lidars are unsightly and costly. Autopilot's hardware systems are largely invisible; even the latest generation of GM/Cruise's Lidar-equipped Bolts have an awkward self-driving rack on top of the vehicle.
Strobe's technology, according to Schoenfeld, allows for lower cost and miniaturization. Think of it like Tony Stark's suits from the "Iron Man" movies: Stark's genius is in making stuff tiny yet powerful. His suits contain the firepower of an F-18 fighter jet and the energy of a nuclear reactor, yet they're physically no bigger than actor Robert Downey, Jr.
GM clearly sees this as a unique opportunity to take Lidar to the next level. And the next level is coming soon.
It helped that Schoenfeld and her partners saw GM as a potential partner early on, back in 2014 when they company had raised only a seen round of funding (it also took some private money, as well as an investment from GM Venture, the carmaker's VC arm that seeks to fund marketable new technologies).
"Our first contact outside the company was with a GM engineer," she said. "They guided us."
But the speed at which GM and Cruise moved, Schoenfeld said, "Stunned me." One the Detroit giant was ready to make the acquisition, Strobe had a term sheet in 24 hours. "It's a Harvard Business Review case study that hasn't been published yet," Schoenfeld said.
"I can't believe my eyes," she added. "They're letting Cruise flourish."