Autonomous driving is approaching a significant milestone: the launch of Waymo’s paid robotaxi service in Phoenix. A recent Forbes article by Alan Ohnsman provides insight into the launch through an interview with Dmitri Dolgov, Waymo’s CTO and vice president of engineering.
We recommend the article as a good read, and agree with many of the points made by Dmitri. As a leader in mapping technology for autonomous driving, we would like to expand on some of Dmitri’s observations.
Dmitri Dolgov, Waymo: “At a high level, there are maybe three things that have to come together. You have to build maps and maintain them…. You have to build and design self-driving hardware…, and then you have to build and be able to deploy the full software stack.”
DeepMap comment: In the self-driving era, a map is no longer ‘just a map’. It’s an integral part of the car’s brain. The autonomous vehicle (AV) doesn’t need a map in the traditional sense. It needs an efficient and reliable answer to questions such as where am I?; what’s around me?; what can I legally and safely do?; and how do I get from A to B? Solving the mapping challenge is at the core of safe autonomy.
Dmitri Dolgov, Waymo: “There’s a lot of stuff that happens at the back end. Data collection is just a part of it. You need the infrastructure, you need the pipelines for crunching all of the data to produce your maps.”
DeepMap comment: For a self-driving fleet to work efficiently, a highly-reliable infrastructure needs to exist for not just building the maps, but also updating the maps. That infrastructure has to support massive data collection, data processing, and machine learning, and do it efficiently (at a low cost), quickly, securely, and reliably. The infrastructure needs to be massively scalable and deployable with different cloud providers, and support various country-specific requirements. It also needs to support various network connections or no network connections at all. DeepMap is demonstrating this capability with customers in Asia, North America, and Europe.
Dmitri Dolgov, Waymo: “It’s not a mature industry. You can’t just go and buy something off the shelf….”
DeepMap comment: Thanks in part to Waymo’s efforts, the AV industry is evolving rapidly. As we wrote in an earlier post, the center of gravity in the self-driving space is shifting from traditional hardware providers to a growing ecosystem of tech players and startups.
In this new world, car makers are faced not only with critical technology decisions, but with new business model and alliance options. Develop in-house? Buy from third parties? Acquire? Acqui-hire? Share? Over the next three to five years, we believe the self-driving industry will adopt a blend of options, on a case-by-case basis.
As the cars of the future become defined more by the software they’re running than by their body type, integrating best-of-class software will be vital to creating great AV experiences. This is especially true for maps for autonomous driving, where we believe the benefits of solving the problem once and sharing that expertise are clear. Through integration, a car maker minimizes cost while maximizing scale, safety, efficiency, and time-to-market.
DeepMap’s mission is to build the world’s best high-definition mapping and localization services in order to accelerate our customers’ AV programs. Safe autonomy is at the core of all our design decisions.
We focus deeply on the precision and accuracy of the maps we build, rigorously optimize on the performance and robustness of our localization service, and stress test our system under the most challenging conditions.