The California Department of Motor Vehicles this week granted Nuro approval to test its third-generation R3 autonomous delivery vehicle in four Bay Area cities, giving the AV startup a positive boost after facing some setbacks and financial struggles.
The approval gives Nuro the ability to test its driverless delivery vehicle in Mountain View, Palo Alto, Los Altos and Menlo Park. Nuro’s vehicles, which don’t have seats, windows, steering wheels or pedals, aren’t designed to carry passengers, only goods. Despite the fact that they operate on public roads, they look more like large sidewalk delivery robots, complete with temperature-controlled storage units to hold food.
The upgraded geographic area will represent the third largest — if not the second largest — deployment of fully driverless vehicles in the United States, after Waymo, co-founder Dave Ferguson told TechCrunch, noting Cruise might have had a larger deployment span before it grounded its fleet late last year.
Nuro also has a 10-year commercial deal with Uber Eats that it’s been testing with third-party vehicles.
Nuro has been teasing its R3 for a couple of years now, but last year decided to pause a planned manufacturing push that would have seen it churn out thousands of vehicles in partnership with Chinese electric car maker BYD. The startup — once a darling of the AV industry after raising over $2 billion from high-profile investors — was burning cash fast. After two rounds of layoffs over the last two years, Nuro restructured its team to focus on getting the autonomy piece right. That meant putting vehicle manufacturing and commercial operations on the backburner.
Ferguson told TechCrunch that Nuro still has no immediate plans to restart scaled manufacturing or heavy commercial operations. The company remains hyper-focused on testing and validating its new AI architecture, and Ferguson says that approach is starting to pay dividends.
“We’ve actually dramatically accelerated our autonomy progress and even the timeline around the autonomy side,” Ferguson said. “So that is the software, obviously, as well as the hardware, the sensing, the compute that’s tied to that autonomy software in a [Level 4] setting.”
The SAE defines Level 4 autonomy as being capable of driving itself without human intervention in certain circumstances.
Ferguson added that Nuro has been testing and validating the R3’s new hardware and software stack on a fleet of retrofitted Toyota Priuses (about 100 according to someone familiar with the matter), and has even continued to do some deliveries with those test vehicles for Uber Eats. In 2022, Uber Eats and Nuro kicked off a 10-year commercial partnership.
Despite putting the BYD manufacturing agreement on hold, Nuro still managed to snag a few dozen R3s from the EV-maker. Within the next few months, Nuro will roll out that fleet in the Bay Area, as well as in its other market of Houston.
A spokesperson for Uber told TechCrunch that the ride-hail and delivery giant expects to start using the R3 for deliveries this fall.
“One of the benefits that the R3 provides, relative to the R2, is that it can go on a significantly expanded [operational design domain],” said Ferguson. “The R2 only drives up to 25 miles per hour. The R3 will technically be able to drive up 45 miles per hour. We won’t necessarily deploy it at that speed on day one, but it enables us to do full L4 driverless testing, deployments, even commercialization over a much wider region, basically everything except freeways.”
Improvements in AI, both at the company and industry level, have helped Nuro make that push. Ferguson said over the past few years, Nuro’s approach has evolved to use one to two very large foundational AI models that perform many tasks – like mapping, localization, perception, prediction and planning – in one place, leading to improved performance and efficiency. Nuro then combines this with a more traditional system, where all those tasks are performed on their own AI models, to validate its AI in real-time.
This not only means that Nuro’s R3 can drive faster and across larger swathes of the Bay Area and Houston, but it also sets the stage for Nuro to scale when it’s ready to do so.
That won’t happen this year, and when it does, Nuro might need to find a new manufacturing partner since anything made by BYD will likely be subject to steep tariffs . Ferguson said the tariffs are a potential concern, but that he’s happy overall with BYD as a manufacturing partner.
In the meantime, Nuro will continue to keep its head down and work on making sure the technology is right and that it’s getting the most out of its Uber Eats deliveries. Ferguson also noted that Nuro is exploring a path to market outside of autonomous delivery, but declined to share more details.