Decoding Nuro: From Last-Mile Delivery to Scaled Autonomy

July 07, 2025.

Nuro is on a mission to enable safe and scalable deployment of self-driving vehicles. What started as an autonomous platform to deliver groceries to consumers has now evolved into a full-fledged licensable self-driving system that has tremendous potential to bring autonomy to existing vehicles. So how did Nuro go from an autonomous goods delivery service provider in California to a full-fledged self-driving car company?

To better answer this question, we need to time travel to 2016 and understand how Nuro has evolved over the years. A custom-built vehicle, half the width and one-third the weight of the average car you see on the road, with no driver, that’s Nuro. That's how you would define it if it was 2016. Founded by technologists from Waymo’s autonomous driving team, Jiajun Zhu and Dave Ferguson, around 2016, Nuro launched its first autonomous delivery vehicle to market in 2018, backed by funding from Greylock Partners and SoftBank Group.

From 2018 to 2020, Nuro partnered with a range of customers to showcase its capabilities: delivering pizzas with Domino’s, prescription medications with CVS, and groceries with Kroger. In 2022, Uber and Nuro announced a 10-year partnership to bring full autonomy to Uber's delivery services.  Now fast forward to 2025, Nuro has developed a top-tier self-driving system capable of bringing autonomy to existing vehicles.

Given its rapid rise in the industry and its recent transition from an autonomous goods delivery service provider to one of the key providers of the modern autonomy platform for existing vehicles, it’s time to understand the vision of Nuro.

Nuro's Vision: From Concept to Reality

Nuro started with a simple yet bold idea: What if robots could drive local commerce and enable shop owners to distribute their goods and services in an autonomous manner? That would allow the local businesses to expand their footprint in a more cost and resource efficient manner by delegating the distribution service to autonomous delivery service providers like Nuro. This approach would strengthen local communities across the country. It's this idea that set the foundation for designing Nuro's purpose-built autonomous vehicles dedicated to transporting goods safely and efficiently within communities. 

The team’s early efforts focused on designing and building compact electric vehicles dedicated solely to enabling last-mile delivery of goods. Over time, Nuro’s vehicles evolved through several iterations, each improving on safety, efficiency, and reliability. The latest generation in this category features spacious, temperature-controlled compartments that keep groceries, dairy products, meals, and fresh produce in optimal condition during transit. If you were to trace the evolution of Nuro's self-driving platform step by step, here is how you would describe it.

It started with P2, a Toyota Prius retrofitted with a rooftop LiDAR sensor. This first prototype of Nuro wasn’t a custom-built vehicle but a conventional car adapted to test autonomous delivery service in Palo Alto, California. This P2 fleet was used to launch pilot programs with major partners, including Kroger for grocery deliveries and convenience stores like 7-Eleven. This first prototype demonstrated  that self-driving vehicles are capable of reliably handling last-mile logistics.

Next came R1, Nuro’s first vehicle designed from the ground up for autonomous delivery. Developed in partnership with Chinese manufacturer BYD, R1 was a small electric vehicle equipped with custom-designed insulated compartments for carrying goods and fresh produce. This was a novel approach adopted by Nuro as this signalled the shift from repurposing existing passenger vehicles to building custom vehicles from ground-up solely dedicated to goods delivery. 

In 2020, Nuro introduced R2, an improved version of R1. The R2 featured a larger cargo area to accommodate bigger orders, was fully electric with zero emissions, and eliminated traditional driving controls such as the steering wheel and pedals. This made it the first vehicle in the United States approved by regulators to operate without those controls on public roads.

By 2022, Nuro unveiled R3, designed for automotive-grade manufacturing at scale. R3 included modular storage compartments that could be configured to carry different types of goods such as groceries, freshly-cooked meals, and packages. It also featured innovative safety capabilities such as an external airbag solely dedicated to help protect pedestrians in the event of a collision. Today, Nuro's vision has evolved from providing a dedicated autonomous grocery delivery platform to a licensable self-driving system capable of enabling autonomy in all modern vehicles. You might ask: What makes Nuro’s approach and design novel?

Nuro's Design Philosophy

While the core design philosophy of Nuro revolves around safety of both the passengers and the nearby pedestrians and vehicles, if you trace the philosophy back to the original design intended for goods delivery, there are a series of novel elements that Nuro introduced with a mission of enabling trust and transfer with end users. Hence its worth understanding Nuro's design philosophy by keeping in mind both the original design intended for goods delivery and the new design intended to bring autonomy to existing vehicles.

If you look closer, here is how Nuro brings this philosophy to life. The first core element that you will notice in Nuro's philosophy is the thoughtful design. If you look closely, the front view of the vehicle resembles the helmet of a racecar driver. This design serves two goals. The first goal here is to ensure that the design exudes a sense of warmth and familiarity that we humans perceive when we look at the face of other humans. The second goal here is to ensure the vehicle is indeed perceived as safe and friendly by pedestrians, bicyclists and nearby vehicles. Now you might ask, what if a pedestrian or a cyclist runs into Nuro? What if Nuro's system is not at fault? Well, that's where the second core element comes into picture. 

If you look at the following graphic carefully, you'll notice that as soon as the vehicle comes into close proximity to a pedestrian it pops open the external air bag. These novel external airbags, the first of their kind, are specifically designed to deploy in the event of a collision and they significantly reduce the risk of injury to pedestrians. This element underscores Nuro’s unwavering focus on protecting vulnerable road users and prioritizing their safety over the goods it carries.

The third core element is Nuro’s focus on efficiency and reliability.  The vehicles feature a lightweight and narrow chassis that greatly minimizes its footprint and significantly elevates its ability to maneuver in tight urban neighborhoods. To ensure operational reliability under all conditions, each vehicle is equipped with triple-redundant computers—three independent systems that process sensor data simultaneously. This redundancy means that even if one system fails, the others can immediately take over in real-time, allowing the vehicle to maintain control and come to a safe stop without compromising its mission or the safety of those around it.

The fourth core element is Nuro’s commitment to sustainability and scalability. The fully electric fleet produces zero emissions and operates largely on renewable energy sources. Finally, the R3 model incorporates custom-designed modular compartments tailored to accommodate diverse types of goods namely, groceries, packaged foods, dairy, and freshly cooked meals. With a capacity of up to 500 pounds and a maximum cruising speed of 45 mph, Nuro vehicles are optimized for efficient, reliable, and scalable delivery operations that integrate seamlessly into modern urban communities. Now that we understand Nuro's approach to design and safety,  it’s time to decode the underlying self-driving system.

 Nuro’s Self-Driving System

Nuro’s self-driving system is a carefully orchestrated blend of advanced technologies that work together  to enable  safe, efficient, and reliable autonomous navigation.  If you’ve read our essay Automotive Is Now Autonomous, you probably remember that the self-driving module of these vehicles boils down to four core pillars—perception, localization, path planning, and motion control. We covered these four pillars in detail in that essay. While these pillars form the core foundation of all modern autonomous vehicles what really makes a difference is the way they are implemented. We will now cover Nuro's approach one module at a time. 

1. Mapping and Data Collection

When Nuro prepares to launch its autonomous delivery service in a new neighborhood, the process begins long before the first delivery vehicle takes to the streets. It starts with building a detailed 3D map of the environment that will serve as the foundation for every decision the self-driving system makes.

This first stage is also referred to as mapping. Nuro’s dedicated mapping fleet which primarily consists of Toyota Priuses retrofitted with high-definition LIDAR sensors and cameras drives through target areas and collects extensive real-world data about the road network, lane features, traffic signals, signage, curbs, and landmarks. Once collected, this raw information is processed through machine learning algorithms and reviewed by human experts to create a high-definition (HD) 3D map. These HD 3D maps serve as the static reference layer that the vehicle relies on to understand its environment before it even starts navigating in this new environment. These HD maps are often updated to include newly added road features or signage that are critical to the vehicle's self-driving system. 

Now, even though the Nuro vehicle has this detailed 3D map, that alone is not enough for it to navigate safely in the urban environments. Why? Modern cities and corresponding neighborhoods are anything but static. Construction zones, emergency vehicles, erratic pedestrians and pokemon-go inspired drivers, there are a lot of variables that demand real-time perception. That's where the next system comes into play. 

2. Sensors & Perception

Once the mapping phase is complete, the vehicles rely on the perception system to understand the behavior and intentions of nearby agents in real-time. This includes studying and analyzing the behavior and trajectories of pedestrians, cyclists, and nearby drivers. How does it do that? That is where the sensors come into play. 

 A custom sensor suite composed of LIDAR, radar, cameras, and thermal sensors continuously scans the environment, constructing a dynamic 3D model of the surroundings. This system identifies and classifies objects such as pedestrians, cyclists, other vehicles, and unexpected obstacles, using advanced algorithms like YOLO (You Only Look Once) and SSD (Single Shot MultiBox Detector). Once the system has identified and classified nearby agents in the scene, the perception layer also estimates the speed and direction of each of those agents. Its goal is to predict how the scene around the vehicle will evolve in the next few seconds so it can make safe and reliable navigational decisions.  

The challenge here is that in addition to the static 3D map and predictions of the trajectories of the nearby agents,  the Nuro vehicle also has to accurately calculate and monitor its exact position on the street. Some of you might say, well, let’s just use Apple or Google Maps. Can we really rely on those maps? Nope. Not in the case of autonomous vehicles. The reason is simple: those maps are only accurate to within a few meters. In self-driving systems, we need to know and track our position with centimeter-level accuracy. We cannot afford to be off by a meter, that could mean the difference between life and death in this domain. 

Why? Because if you are off by a meter in predicting your position, technically your car could think it’s in a completely different lane than where it actually is. That means any navigation decision it makes based on that inaccurate position could lead to serious, even catastrophic, consequences. This is why a high-precision localization system is absolutely critical. And that’s where the next step module comes into play.

3. Localization: Tracking the Vehicle’s Position

While the perception module helps the vehicle develop a coherent understanding of its environment and predict how the scene around it will evolve over next few seconds and minutes, the localization system ensures the vehicle is able to identify and track its precision location at all times. How does it do that? 

First, the vehicle tracks its approximate position using a GPS. Then it fuses the data received from the sensors in real-time and correlates this dynamically generated scene of its environment with the HD map that was fed to it during the training sessions. To compare and identify the exact position on that HD map, these vehicles typically use map-matching algorithms. We cover this approach in great detail in our course on self-driving cars. These algorithms allow the Nuro vehicle to track its position with centimeter-level accuracy at all times. With a clear understanding of the environment and its exact position, the vehicle is now in a state to make reliable navigational decisions. How does it do that? That’s where the next system comes into play.

4. Path Planning and Prediction

The path planning system relies on artificial neural networks that are trained on extensive datasets collected from diverse scenarios encountered by a fleet of test vehicles deployed in each city before launching the production vehicles. These neural networks are trained on real-world data collected from test fleets and 1000s of scenarios that are simulated by the engineering teams to ensure the vehicle is fully capable of handling a wide range of agents as it navigates across the cities. 

When on the road, the Nuro vehicle uses this neural network to assess a wide range of route options and filter out the one that meets critical criteria: it must be safe, legal, comfortable, and efficient. Before deploying these path planning algorithms to production vehicles, they are rigorously tested in simulated environments. These virtual simulations replicate real-world driving conditions, allowing the system to encounter and respond to a wide range of scenarios with minimal risk. By running millions of simulated miles, Nuro ensures its artificial neural network driving the navigation decisions exhibits mission-critical reliability. Now the path planning system provides a trajectory or a route that the Nuro vehicle has to implement. That route should be translated into corresponding action. How does it do that? That's where the next system comes into play. 

5. Motion Control System

Once a safe and efficient trajectory has been selected, the motion control system executes the navigation plan by translating it into precise commands for steering, accelerator, and brakes. This layer is responsible for ensuring that the vehicle follows the path finalized by the path planning system with highest degrees of precision and reliability.  In rare instances where the vehicle encounters an unexpected scenario it may not be trained to handle it will send an alert to a remote operator. The remote operations team will take the controls, help navigate the vehicle through challenging scenarios and ensure safe and reliable operations. Now you might ask: Who are the customers of Nuro? Who is actually using these systems? 

Partnerships & Road Ahead

There are a series of customers that find value in Nuro's mission and offering. From Domino's to FedEx and UberEats to CVS,  these partnerships have demonstrated Nuro’s potential to revolutionize last-mile delivery. Over the years, the vision of Nuro has evolved. The company has now pivoted from providing its own autonomous delivery service to enabling existing automotive manufacturers to add autonomy to their vehicles at scale.

With a fresh round of $106 million in Series E funding, Nuro is expanding its offering both in the U.S. and in Japan. Right now, pilot tests are underway across the United States, including but not limited to Memphis, Las Vegas, and Houston. Given this rapid expansion by companies like Nuro, Waymo, Tesla, and others in the autonomous vehicle domain, it’s important to understand the core principles and the underlying technologies these companies have developed with depth and clarity. 

We cover these systems in great detail in our Self-Driving Cars course. In this course, you'll learn how to build a self-driving system from the ground up. You'll gain a deep understanding of the autonomous stack developed by Tesla, Waymo, NVIDIA, and Zoox. Our courses have helped over 80,000 students from 150 countries kick-start their careers in these emerging fields.

If you're serious about entering the self-driving car industry, this course is for you. 

Cheers,

Prathamesh

Disclaimer: This blog is for educational purposes only and does not constitute financial, business, or legal advice. The experiences shared are based on past events. All opinions expressed are those of the author and do not represent the views of any mentioned companies. Readers are solely responsible for conducting their own due diligence and should seek professional legal or financial advice tailored to their specific circumstances. The author and publisher make no representations or warranties regarding the accuracy of the content and expressly disclaim any liability for decisions made or actions taken based on this blog.