Nvidia investing $10M in last-mile company Serve Robotics

Listen to this article

Serve Robotics uses autonomous mobile robots to deliver items to customers, while sharing the sidewalk with pedestrians. | Credit: Serve Robotics

Nvidia is investing $10M into last-mile delivery company Serve Robotics. Serve Robotics spun out of Uber in March 2021 after being acquired under prior company name Postmates by Uber in November, 2020.

Serve built its sidewalk delivery solution around the Nvidia Jetson edge compute module. Data from all of the onboard sensors and cameras are processed by the Jetson Xavier computer. This information is leveraged for obstacle avoidance, navigation and localization.

In December 2021, Serve Robotics raised $13 million in an expanded seed round, bringing on investors to help the company on its path toward commercialization. This included Delivery Hero-backed DX Ventures, 7-Eleven’s corporate VC arm 7-Ventures, and Uber. In this latest investment, Nvidia adds an addition $10M to the mix.

Here’s a recent TED Talk by Ali where he explains the service:

I caught up with Serve Robotics co-founder and CEO Ali Kashani to ask him about the relationship with Nvidia and the current state of the company.

Mike Oitzman (The Robot Report): What are some of the challenges that Serve Robotics has had to solve with this specific application for mobile robots?

Ali Kashani (Serve Robotics): The mantra in the team was always “the truth lies on the sidewalk”. So we needed to go out and figure out all of the edge cases and things we needed to think about with this robot.

One of the things that we observed early on was that people on the street don’t always see the robots. Their attention is often on their phone and the robot is small enough that pedestrians can trip over it if they don’t see it. So we had to make sure that the robot design had the right volume and size to be visible in the crowd on a sidewalk.

One of the other things about vehicles of any kind, is that stopping distance is a measure of safety. You want it to stop fast to avoid an obstacle or collision, but we had to increase the stopping distance. This is because if the robot stops too fast in a crowd then it can  trip a human walking behind the robot. The robot becomes a hazard.

So you have really interesting dynamics that you don’t think about until you’re actually outside and interacting with the environment.

As a startup, the other big challenge was the pandemic. We didn’t know how that would impact our business. With the restaurant and delivery labor issues, we didn’t know how that would impact us. But, it turns out, it was another tailwind for the industry. So it’s been a  very wild ride.

Since you’ve spun out of Uber, does Uber Eats remain your primary partner? Or are you looking for other partners? What’s the roadmap look like for you going forward?

The primary reason we wanted to become an independent company was so that we can partner with others. And that’s exactly what we’ve done since leaving Uber. Uber is a very important partner. But we are also able to work with other folks.

In December we also took investment from 7/11 and Delivery Hero. There are other partnerships in the pipeline that we are not able to talk about just yet, but I hope we will be able to soon.

Can you expand a little bit on the Nvidia investment? You’ve engineered the Nvidia solution into the heart of the robot. Tell us why why the relationship and investment with Nvidia is good for both of you?

With Nvidia, specifically, this was a very unique type of investment and collaboration. Our relationship with Nvidia goes back all the way to 2017 when we first started working with them. They’ve been a close partner. We worked hand in hand with them as we were developing our product, and as they were developing some of their product lines.

So you work closely with the Nvidia product management and development teams to prioritize features that support your needs?

That’s right. Within the Nvidia Jetson line, there’s a number of things we’ve been interested in, have worked with already, or want to work with down the line. For our engineering teams it’s kind of a badge of honor for us too, because Nvidia has worked with a number of companies in the space.

So this is a validation of how they see the space and where they think we are in terms of the technology. Again, we believe we are leading the way when it comes to the core technology.

Serve Robotics is currently offering service in LA and in San Francisco. What’s next on your roadmap?

We offer service in the San Francisco bay area. We are building out our fleet right now, and the influx of capital helps to achieve that goal.

We are preparing to launch in more locations, both in places where we currently operate and in new markets. So by the end of this year, I expect you will see us in many other places as well.

The early market looks like it’s urban areas with well-defined sidewalks. Do you need well structured sidewalks and other infrastructure to function?

Something like 70% of the country has sidewalks, so we kind of rely on that. The infrastructure doesn’t have to be perfect. I don’t know if you’ve been in LA, but the sidewalks are not always in great shape there sometimes. But that’s what the robots are designed for.

Generally speaking, we like population density, but it doesn’t have to be as dense as Manhattan or downtown San Francisco to be viable.

Ideally, we like to start on a few streets where a lot of restaurants are congregated. That becomes a good base of operation because usually within a couple of miles, there is significant population density that we can deliver to.