Aurora's technology and strategy are sound. As it prepares for an IPO, the AV developer next needs to execute on its roadmap.
This post looks at Aurora Innovations’ autonomous vehicle activities with perspectives on the self-driving specialist’s strategy, business model and technology. Aurora is in the midst of a special purpose acquisition company IPO scheduled to be completed in early November.
We’ve drawn on Aurora’s investor presentation and the company’s website. The table below summarizes Aurora’s AV development activities.
Aurora Innovation is a leading developer of a software platform for autonomous vehicles. It was founded in 2017 by three pioneers in the AV industry. Chris Urmson, co-founder and CEO, has extensive experience from Google-Waymo and participated in the DARPA challenges. Sterling Anderson, Aurora’s co-founder and chief product officer, worked on Tesla’s autopilot and related AV development. CTO Drew Bagnell is an associate professor at Carnegie Mellon University. Bagnell has extensive experience in AI, machine learning and related fields.
Aurora has more than 1,600 employees, including about 175 PhDs, and more than 1,400 product developers and engineers.
Aurora agreed to go public via a merger with Reinvent Technology Partners, a SPAC led by LinkedIn co-founder Reid Hoffman and Zynga founder Mark Pincus. The implied valuation for Aurora is $13 billion. Aurora is expected to raise $2 billion from the offering, including cash held in Reinvent’s trust and $1 billion in PIPE funding. The PIPE investment originated from Baillie Gifford, Fidelity Management & Research, T. Rowe Price, Counterpoint Global (Morgan Stanley), the Canada Pension Plan Investment Board, Primecap Management, Reinvent Capital, XN, Index Ventures and Sequoia Capital.
Strategic investors include Uber, Paccar and Volvo Group.
Aurora has so far raised $1.22 billion in venture funding, including a $400 million investment from Uber as part of Aurora’s acquisition of Uber’s self-driving vehicle unit.
The SPAC and earlier VC investments will provide Aurora with about $2.5 billion at closing. The funding and previous investments are expected to provide Aurora with enough capital to last through deployment of autonomous trucks and early stages of robotaxis and, perhaps, goods delivery AVs.
Strategy, business models
The slide below from Aurora’s investor presentation provides perspectives on its development timeline and product strategy. It shows five phases from 2017 to 2025 and beyond. For each phase, Aurora includes information on its products, software driver, vehicle use cases and operational aspects.
The Aurora Driver is being developed for multiple AV use cases. Autonomous trucking is the first, with hub-to-hub trucking as the starting point. There are major advantages in starting with autonomous trucks.
Aurora’s second segment is ride-hailing with AVs, which include robotaxis and mobility services.
Aurora provides its AV technology to an external fleet owner or operator and charges a per-mile fee. Aurora calls this Driver-as-a-Service, meaning its revenue stream is based on miles driven by customers.
Aurora covers variable costs such as insurance, Aurora Driver hardware and maintenance expenses, tele-assist, cloud processing, telecommunications and any variable fees paid to partners. It also covers fixed costs such as development and extension of Aurora Driver.
Tele-assist is similar to teleoperation for remote interactions when needed.
Autonomous truck fleets using Aurora technology will be owned and operated by third parties.
Aurora plans to deploy an autonomous trucking business by late 2023 through partnerships with truck manufacturers Paccar and Volvo Trucks along with logistics companies such as FedEx.
The Volvo VNL truck is the first prototype using the Aurora Driver as an autonomous software platform, and is intended for commercial production. On-road testing of the Aurora Driver-based VNL will begin in 2022.
Aurora plans to deploy ride-hailing in late 2024 with companies such as Uber and Toyota. A Toyota Sienna equipped with Aurora Driver will serve as the prototype for Toyota’s first mobility services vehicle, with future volume production planned. In September, a fleet of around 12 Toyota Siennas with Aurora Driver started testing in Pittsburgh, San Francisco Bay Area and Texas.
Another investor presentation slide shows Aurora positioning its technology as “Self-Driving 2.0” for autonomous vehicle use cases. The startup is advancing its AV technology through learning and improvements based on previous development experience.
Aurora is also transitioning to AV sensors 2.0 for lidar, radar and cameras. In 2019, it acquired Blackmore, a startup providing frequency-modulated, continuous-wave (FMCW) lidar technology, a vast improvement over traditional lidar. Aurora’s FirstLight FMCW lidar is now in use.
The FMCW lidar enables quicker reaction and longer range (400 meters) with speed estimates for moving objects based on the Doppler effect. It also eliminates virtually all interference from sunlight and other sensors.
Aurora has also added imaging radar and custom cameras where needed for better sensor capabilities and safety.
As with most other AV software developers, Aurora is using Nvidia Drive as its computing platform. Aurora’s latest hardware platform was announced in August, and includes hardware redundancy.
Aurora’s Virtual Testing Suite includes driving simulation and data replay technologies with multiple benefits such as cost savings and accelerated testing. Aurora’s motion planning simulation is billed as 2,500 times less expensive than on-road testing. In one hour, the testing suite can simulate the equivalent of more than 50,000 trucks operating on the road.
Better testing safety is another plus. Aurora was able to simulate 2.25 million unprotected left turns before testing on public roads. That should greatly reduce the number of on-road miles needed to develop Aurora Driver, especially learning and improving edge cases.
Aurora Atlas is a HD mapping system. The Atlas architecture provides accuracy where it’s needed most: near the vehicle, including near-real time updates. The comprehensive map data structure provides rapid and parallel map building, allowing rapid updates in the field.
Aurora has so far filed or received more than 1,100 patents and patent applications.
The focus is on providing its software driver platform as a service to variety of AV use cases. Three services are planned, according to the company’s website:
Aurora has identified strong partners, some who are also investors. In the trucking industry Aurora has two leading truck OEM partners—Paccar and Volvo Trucks. It also has a trial with FedEx, which is among the top logistics players. More logistics companies are expected to become partners.
In the mobility services segment, Aurora is also faring well—Toyota and Uber are partners and investors. Toyota and Aurora have a long-term partnership commitment. Toyota is also a large investor and major development partner. Denso, Toyota’s key Tier 1 supplier is also a partner. Denso is a potential manufacturer of Aurora’s hardware and sensor platforms.
Aurora and Uber established a long-term commitment included as part of Aurora’s acquisition of Uber’s AV development group in December 2020. Uber became a large minority investor, and its CEO is on Aurora’s board.
A 10-year agreement gives Aurora access to Uber data from its ride-hailing operation. The Uber data is valuable for training and testing the Aurora Driver, along with many other benefits.
Detailed market data also enables Aurora to select the best robotaxi markets for AV training. Data on ride-hailing trips and what roads are most used would allow it to prioritize software driver capability development.
AV testing, plans
Aurora has focused on gaining virtual testing experience before launching road-miles testing. Its robotaxi road miles are primarily in Mountain View, San Francisco and Pittsburgh. It will soon start robotaxi road testing in Dallas.
Its truck road testing started in Texas in September with partners Paccar and FedEx. Autonomous truck testing initially includes safety drivers. The tests utilize FedEx hub-to-hub round trips between Dallas and Houston, which is a 500-mile roundtrip. Over the next few years, autonomous trucks will primarily operate on interstate highways in Texas, New Mexico and Arizona. Deployments may change depending on its future logistics clients.
Aurora’s business plan projects autonomous trucks will drive 1 million miles in 2023, which increases to 20 million miles in 2024 and 134 million by 2025. Aurora’s yearly autonomous truck miles will jump to 3.25 billion by 2027.
The table below summarizes Aurora’s revenue estimates. It expects to continue generating partner development and pre-commercialization revenue with vehicle operators before launch in 2023.
It also anticipates revenue from trucks without safety drivers in late 2023, with a small fleet of 20 trucks owned and operated by Aurora.
By adding projected vehicle miles traveled for trucks and rides, estimated revenue per mile for Aurora was calculated in the table above. It’s clear from Aurora’s forecast that revenue per mile will decline over time.
Aurora has a good technology plan, and its business model has excellent long-term potential. The timeline looks reasonable, but AV complexity issues could delay Aurora and its competitors.
In terms of safety and planned deployment timeline, Aurora appears more cautious in its AV development approach than many of its competitors.
The Driver-as-a-Service business model is based on revenue per mile, requiring many annual miles to generate a large revenue stream. This could mean revenues will grow slowly at first, requiring Aurora to find other revenue sources.
Business models based on services that leverage software platforms are favored by investors, giving Aurora an opportunity if it executes on its plan.
This article was originally published on EE Times.
Egil Juliussen has over 35 years’ experience in the high-tech and automotive industries. Most recently he was director of research at the automotive technology group of IHS Markit. His latest research was focused on autonomous vehicles and mobility-as-a-service. He was co-founder of Telematics Research Group, which was acquired by iSuppli (IHS acquired iSuppli in 2010); before that he co-founded Future Computing and Computer Industry Almanac. Previously, Dr. Juliussen was with Texas Instruments where he was a strategic and product planner for microprocessors and PCs. He is the author of over 700 papers, reports and conference presentations. He received B.S., M.S., and Ph.D. degrees in electrical engineering from Purdue University, and is a member of SAE and IEEE.