Why Cruise is making its own chips, and a lot more besides • TechCrunch
Cruises never plan to create their own silicon. But in the quest to commercialize robotaxis – and make money from it – those never-planned pursuits suddenly seem a lot more appealing.
Cruise realized that chip prices from suppliers were too high, the parts were too large, and the reliability of third-party technology wasn’t there, Carl Jenkins, Cruise’s vice president of hardware, told TechCrunch in a statement. company tour. hardware lab last month.
Between a hiring spree starting in 2019 and continue in 2020, Cruise has doubled down on its own hardware, including its own boards and sensors. The investment helped the company develop smaller, lower-cost hardware for its vehicles. It also created the first production board, the C5, which powers the current generation of self-propelled Chevy Bolts.
When the company’s purpose-built robotaxi Origin starts hitting the streets in 2023, it will be equipped with a C6 panel. That board will eventually be replaced by a C7 that will feature Cruise’s Dune chip. Dune will process all sensor data for the system, according to Cruise.
Typically, automakers use parts and sensors from Tier 1 suppliers to reduce R&D and manufacturing costs. Cruise couldn’t figure out how to launch its autonomous ride-hailing without doing more work on its own. The result is that C7 boards are 90% cheaper, 70% less mass, and use 60% less power than vendor-supplied chips.
It’s not just the chips that the company is interested in. While long-range lidar and ultrasonic sensors are still sourced from third parties, nearly everything else, including cameras, short-range lidar and radar, is also being developed in-house.
Cruise discovered that the unavailable radar did not have the necessary resolution for their vehicle to function. Like the board, long-term costs will be reduced by about 90%, according to Jenkins.
“I was told how much I had to pay for this hardware for 2025,” says Jenkins. “So I went to see all the Bosch, Continental and ZF CTOs in Germany. ‘Do you have anything in your research tanks that you’re working on that meets this?’ Nothing, not even starting. ‘Okay, if you start today, how long should I take?’ Seven years.”
At that point, Jenkins was able to grow his 20-man team to 550.
When asked about the cost of building Origin with in-house developed hardware compared to parts sourced from vendors, CEO Kyle Vogt told TechCrunch, “we couldn’t do that. It does not exist. “
However, that doesn’t mean Cruise doesn’t want to buy the hardware it needs.
“What we’ve found in the AV industry is that a lot of the components that have the durability needed to operate in harsh automotive environments don’t have the capabilities needed for an AV. Vogt says the necessary AV-capable components cannot operate in those harsh environments.
Made at Cruise, used at GM?
Automakers (not counting Tesla) have taken a more cautious approach to autonomous vehicles that will be sold to consumers. Technology built and proven by Cruise could eventually become a product GM sells to customers.
And there’s reason to believe it will.
GM CEO and President Mary Barra has repeatedly said that the automaker will Build and sell personal autonomous vehicles mid-decade.
GM President Mark Reuss told TechCrunch editor Kirsten Korosec in a recent interview: “We use Cruise as a tether for autonomous vehicle technology and stacking and how it works. . When Cruise developed AV technology, its parent company focused its efforts on the advanced driver assistance system Super Cruise and now Ultra Cruise.
“When we start researching and looking at personal autonomous vehicles, there will be options like the bike has pedals or it has deployable pedals or it has no pedals at all.” Reuss said. “And so we’re looking at what people want and those aren’t easy questions to answer.”
With just a few years to go before its mid-decade goal, GM still has a lot of work to do, including a go-to-market strategy for these personal autonomous vehicles (or Reuss calls them PAVs). Its recent responses InnerSpace autonomous concept for Cadillac
GM has yet to decide whether these PAVs will launch as a marketable product or if it will be attached to an existing model or a dedicated vehicle, Ruess added.
Troubles on the road
Cruise currently runs an autonomous ride-hailing business in San Francisco but operates only at midnight (10 p.m. to 5:30 a.m.) and within 30% of the city. The company notes that the decision is more based on ensuring its vehicles operate during less busy traffic times. It is currently working to extend the space and time constraints.
It’s not just San Francisco that will see more driverless Chevy Bolts carrying passengers around. Cruise plans to expand to Phoenix, Arizona and Austin, Texas in the next 90 days.
Scaling is the next chapter of Cruise. However, the hiccups kept coming. There have been numerous reports of cruise control robots blocking intersections and other problems.
A vehicle was involved in a collision at an intersection that caused the company to Update software on 80 vehicles of the company. In April this year, a Bolt was towed over for not having its headlights on and at one point was towed away from police. And of course, there is a notorious group of over half a dozen Travel bolts assembled at an intersection and cannot determine where to go next causing traffic problems.
When asked about assembling vehicles, Vogt noted, “It’s part of the operation, part of scaling. It was a normal crash on the road. “The CEO noted that it was an inconvenience and not a safety issue. Vogt said that AV had a lot of back-end services and one of them was “flipped” and didn’t come back online. The reason why they were all at the same intersection was that at the time there was only one launch site for vehicles and they were flying along one of their main corridors near that launch site. Since then Cruise has incorporated restorative techniques in AVs to make them more tolerable.
The company (and by extension Vogt) is confident in its in-house automated ride hailing system. Now it needs to convince skeptics that a ride in a driverless car is worth paying in cities outside of tech-friendly San Francisco.
Our unmanned ride
At the end of the tour, Cruise set us up for an auto ride in a Bolt.
Our car, named Ladybug, arrived and with a single tap of the app we unlocked the doors and flew (no pun intended) around the city at night on our way there. Japanese town.
Along the route, many vehicles were parked with their driver’s side doors open. The Bolt slowed slightly, flashed, and sped across to the other lane before landing again. At four-way stopping intersections, it takes on the character of a cautious person, stopping only after determining that other vehicles will follow the rules of the road.
It was fun at first and then, boring, which is exactly what driverless ride services should focus on. Yes, it’s a bit odd to be in a robot-driven car, but after 20 minutes of being carefully driven by a robot, the last 10 minutes are spent wondering if you’re stuck at an intersection. just to add some excitement. trip.
Additional reporting from transportation editor Kirsten Korosec.