Aurora’s Software Engineering Tenets
Scaling gracefully is a balancing act. Here’s how we’re building a culture of trust and accountability where we can both push boundaries and build production-ready software.
Today, we announced we are welcoming Uber ATG to the Aurora family.
We are uniquely positioned to transform the world of transportation, and every decision we make as a company is in service of our mission. ATG’s team and technology will be pivotal to delivering the Aurora Driver safely, quickly, and broadly.
As I’ve gotten to know the ATG team, I’ve been continually impressed by their resilience, tenacity, and grit. They are passionate about delivering a product that is scalable, reliable, and safe, and they have demonstrated their ability to strategically grow from experience. In terms of technology, while many of their achievements have flown under the radar, they have repeatedly made impressive strides across the business that will accelerate our work.
Below, I’d like to highlight some of the reasons why ATG is such a valuable addition to Aurora.
ATG led the industry by being the first self-driving company to publicly share a Safety Case framework for self-driving. In addition to robust safety engineering, their approach emphasizes and includes organizational safety, like directly measuring safety risk and whether safety policies and processes are clear, available, and understood by employees. Throughout the last several years, they’ve also worked tirelessly to build an experienced safety team with diverse backgrounds, embed safety processes and standards into the fabric of their day-to-day work, and educate their employees. Finally, they’ve developed a strong systems engineering team who has implemented rigorous multi-level verification and validation, helping to ensure that hardware and software meet both performance and safety requirements at each stage of the development process. Safety is paramount and central at Aurora, and ATG’s experience and commitment to safety will complement the safety engineering work our team has been doing.
Uber's most recent Safety Report can be found here.
Aurora’s Software Engineering Tenets prioritize rapid experimentation that allows development teams to innovate and iterate quickly. ATG has also embraced this approach by emphasizing thorough testing throughout their stack and creating sandboxes where developers can safely run experiments and directly evaluate the impact and benefit of changes. They have also invested heavily in creating a log-data ecosystem—adding metrics, building tools to perform deep log search, etc.—that allows engineers to quickly find the right data for performing experiments, training models, and more. Finally, they’ve created extensible, web-based visualization tools that give employees across the organization insight into the performance of each part of the system and how their code translates to the road. We look forward to building upon this work together and accelerating the velocity of our combined team.
ATG has built excellent hardware for their vehicle fleet. Their 4th-generation kit includes a powerful computer and custom-built cameras designed to optimize navigation in urban environments. Throughout each iteration, ATG’s software and hardware teams have closely collaborated to tackle challenges like detecting and responding to pedestrians walking very close to the vehicle—a common occurrence in busy city intersections. Over the years, they’ve developed several innovative solutions across sensor configuration, testing, and data processing.
Much like our hardware team, ATG has deep knowledge of industry best practices, a proven track record of seeing strong designs all the way through to completion, and a commitment to safety. But we also have a lot we can learn from each other. ATG has incorporated some elements into their hardware kit that were further out on our product roadmap, like liquid cooling for their computer. With the combined perspective from each team, we’ll be even better equipped to create scalable hardware for a wide range of vehicle types.
At Aurora, many of us are building self-driving technology for the second, third, or even fourth time, so we prize the ability to leverage experience to build next-generation systems. We also appreciate the humility required to think deeply about how things can improve and the boldness required to take a new direction. ATG has done this repeatedly throughout their organization, and we look forward to bringing that work ethic to Aurora.
Recently, they completed an impressive rewrite of their entire software platform, making their software significantly more efficient and accelerating their simulation efforts. They also introduced a new framework for building web-based tools (which has served them well while working from home) and completely reworked their process for building detailed, high-quality AV maps. These are just a few examples of how they have repeatedly set and achieved outrageous goals, one of our core values.
While we have been advancing steadily toward our first logistics product, ATG has focused on ridesharing. They’ve been thinking through challenges like rider assistance, dispatching self-driving vehicles alongside human drivers, and nuanced robotics problems such as rider pickup and dropoff. Informed by over 50,000 trips on the Uber platform, they’ve also invested heavily in understanding rider experience to design a product that is not only functional, but also comfortable for passengers. This foundational work and combined expertise gives Aurora an unprecedented opportunity to lead the industry in both trucking and passenger mobility.
Delivering the Aurora Driver broadly requires effectively and efficiently scaling our teams and technology, another area where ATG brings a wealth of experience. First, they’ve hired, trained, and managed large operational teams, skills that will be critical as we continue to grow our delivery and ridesharing fleets. They have also strategically built and refined tooling, workflows, and automation to accelerate efforts like label production, with impressive results. Finally, they’ve invested heavily in cloud infrastructure that allows them to efficiently manage, analyze, and process increasing amounts of complex data.
From their leadership in safety and structured testing, expertise in ridesharing software and scaling, to their passion for changing the world—ATG is an extraordinary team. Self-driving is potentially the most challenging engineering problem of our lifetime, and together we’ll be able to bring the Aurora Driver to more people much faster than either of us could have alone.
Aurora, founded in 2017 by the leading experts in self-driving, is delivering the benefits of self-driving technology safely, quickly, and broadly. The company is building the Aurora Driver, a platform that brings together software, hardware and data services to operate passenger vehicles, light commercial vehicles, and heavy-duty trucks across a range of applications. Aurora is backed by Amazon and Sequoia, among others, and tests its vehicles in the Bay Area, Pittsburgh, and Dallas. The company has offices in those three cities and in Bozeman, Montana.
Visual Assets: https://aurora.tech/press/