Riding on board with Rivian’s race toward autonomy

The robot careened through the cafeteria of Rivian’s office in Palo Alto, where shelves were lined with chilled packaged coffee — until it wasn’t. Five minutes later, a man carefully pushed it out of everyone’s way, and the words “I’m stuck” flashed yellow on the poor robot’s screen.

It was an inauspicious start to Rivian’s “Autonomy and AI Day,” a showcase of the company’s plans to make its cars capable of driving themselves. Rivian didn’t make the cafeteria robot and isn’t responsible for its abilities, but there was a familiar message in its weaknesses: This stuff is tough.

Hours later, as I rode the 2025 R1S SUV during my 15-minute demo of Rivian’s self-described new grand driving model, I remembered that message.

The electric vehicle equipped with automated driving software drove me and two Rivian employees down a winding road near the company’s campus. As we passed by Tesla’s engineering office, I noticed a Model S in front of us slow to turn into the competitor’s lot. The R1S eventually noticed this too, and slammed on the brakes almost before a Rivian employee could intervene.

During my test drive, there was one actual withdrawal. The employee in the driver’s seat took over as we drove through a section of road that was one lane due to some tree trimming. Simple things in general. But it wasn’t exactly rare either; I’ve observed several other test flights that have had disengagements as well.

The rest of the driving went well enough for software that’s not ready to ship, especially when you consider that Rivian ditched its old rules-based driver assistance system and adopted a comprehensive approach — something Tesla developed for full (supervised) self-driving. It stopped at stop signs, negotiated turns, and slowed for speed bumps, all without programmed rules telling it to do those things.

Quiet pivot in 2021

Rivian’s old system “was very deterministic, and it was very structured,” Rivian CEO RJ Scaringe said in an interview Thursday. “Everything the car did was the result of a specific control strategy written by humans.”

Scaring said that when Rivian saw Transformer-based AI taking off in 2021, he quietly “reformed the team and started with a clean sheet and said, let’s design our self-driving platform for an AI-centric world.”

After spending “a lot of time in the basement,” Rivian is launching the new ground-up drive software in 2024 on second-generation R1 vehicles, which use Nvidia’s Orin processors.

Scaring said his company only started seeing significant progress recently “once the data really started rolling in.”

Rivian is betting that it can train its Large Driving Model (LDM) on fleet data so quickly that it will allow the company to roll out what it calls “global hands-free” driving in early 2026. That means Rivian owners will be able to take their hands off the wheel on 3.5 million miles of road in the U.S. and Canada (as long as there are painted lines visible). In the back half of 2026, Rivian will allow “point-to-point” driving, or the consumer version of the demo we received on Thursday.

Challenge from “raising eyes” to “raising hands”.

By the end of 2026, after Rivian begins shipping its smaller, more affordable R2 SUVs, it will ditch Nvidia’s chips and equip those vehicles with a new custom computer that was unveiled Thursday. This computer, combined with the lidar sensor, will eventually allow drivers to take their hands and eyes off the road. True autonomy – where you don’t have to worry about taking back control of the car – lies well beyond that and will largely depend on how quickly Rivian trains its LDM.

This rollout represents a near-term challenge for Rivian. The new standalone computer and lidar won’t be ready until months after the R2 goes on sale. If customers want a car that can handle driving without looking (or more), they’ll have to wait. But the R2 is a critical product for Rivian, and the company needs it to sell well — especially in the wake of declining sales of first-generation vehicles.

“When technology moves this quickly, there’s always going to be some level of obsolescence, so what we want to do here is be really direct” about what’s to come, Scaring said. Early R2 cars will still get Rivian’s promised “point-to-point” driving, which will rely on the new software and be hands-free but without the eyes lifting.

“So [if] “If you buy an R2 and buy it in the first nine months, it’s going to be more restrictive,” he said. “I think what will happen is some customers will say ‘This matters a lot to me, I’ll wait.’ Some will say ‘I want the latest and greatest stuff now, and I’ll get the R2 now, and maybe I’ll trade it in a year or two, and I’ll get the next one later.'” Fortunately, there’s so much backlog of orders for the R2 that we think that by being honest about this, customers can decide for themselves.

“In an ideal world, everything would happen at the same time, but the timeline of the vehicle and the timeline of the self-driving platform are not exactly the same,” he said.

When I first interviewed Scaringe in 2018, even before Rivian showed what its vehicles looked like, he shared a goal that was still on the back of my mind. He wanted to make Rivian vehicles self-driving so that: “If you go for a walk, and you start at one point and end at another, the car will meet you at the end of the path.”

It was the kind of impossible promise about self-driving cars that was all the rage seven years ago, but it stuck with me if only because it was something that seemed true to Rivian’s whole brand of aspirational adventure.

Scaringe told me Thursday that he still believes it’s possible for Rivian to enable a use case like this in the next few years. That certainly won’t happen until the company tests and builds the more capable R2 vehicles, which would take at least a year in the best-case scenario.

“We can [do that]. “That hasn’t been a big focus,” he said. That could change as the company approaches Level 4 autonomy, because by then the company will have trained its LDM on more difficult roads without directional features like lane lines.

“Then, it becomes a bit similar, what is ODD [operational design domain]? Dirt roads, off road? “Easy,” he said. “Just don’t expect a country car to drive itself over Hell’s Gate in Moab.”

“We’re not putting any resources into autonomous rock crawling,” he said. “But as far as getting to the trailhead? Sure.”