Deepmap builds high-definition maps for autonomous vehicles. | Image credit: Nvidia
NVIDIA adds high-resolution mapping capabilities to its autonomous driving arsenal with the announcement this week of its intention to acquire DeepMap. DeepMap, founded in 2016, is a startup dedicated to building high-definition maps for autonomous vehicles to navigate the world safely. Chipmaker Nvidia deepens its technology assets with the acquisition. The deal, for an undisclosed amount, is expected to close in Q3 2021.
Map making precision from DeepMap
Silicon valley based DeepMaps was founded by James Wu and Mark Wheeler, veterans of Google, Apple and Baidu. The company is attempting to resolve the issues of precision with maps for autonomous vehicles. It is also is working on solutions to deliver ultra-accurate real-time localization for autonomous vehicles. Every autonomous vehicle manufacturer is struggling with these two issues.
Autonomous vehicles require centimeter level precision. DeepMaps is working on the mapping technology to support this requirement. They are also developing the infrastructure to deliver maps-as-a-service, so that map information can be continuously updated and delivered to the vehicles. This requires the creation of a cloud-based delivery network for pushing information to hundreds of millions of end points (i.e. vehicles).
Nvidia’s primary product for autonomous driving is its Nvidia Drive platform. As a chip-maker, Nvidia’s ultimate goal is to sell more silicon. However, their processing platforms are unique suited to the tasks supporting artificial intelligence (AI) workflows and machine learning (ML) models. The perception tasks for autonomous driving are some of the primary areas where AI/ML is being applied today, and where current Nvidia hardware is being deployed to supporting the computing requirements.
A DeepMap map in San Jose, Calif., depicting highly detailed features of the road and surrounding city block environment, including a reliable semantic layer of information with key attributes such as navigable boundaries, lane boundaries, crosswalks, traffic signs and traffic signals, explicit and implicit yield lines, and lane connectivity.
“The acquisition is an endorsement of DeepMap’s unique vision, technology and people,” said Ali Kani, vice president and general manager of Automotive at NVIDIA. “DeepMap is expected to extend our mapping products, help us scale worldwide map operations and expand our full self-driving expertise.”
“NVIDIA is an amazing, world-changing company that shares our vision to accelerate safe autonomy,” said James Wu, co-founder and CEO of DeepMap. “Joining forces with NVIDIA will allow our technology to scale more quickly and benefit more people sooner. We look forward to continuing our journey as part of the NVIDIA team.”
Autonomous vehicle localization is a core technology for DeepMap
Mapping and localization go hand in hand for every autonomous vehicle, from unmanned aerial drones to autonomous mobile robots, to autonomous automobiles. DeepMap has invested heavily in the problem of improving the precision of AV localization. To localize more precisely, you need more accurate maps. Conversely, with more accurate maps, you can more easily and precisely locate something. This is the reason why both of these technology areas are a part of DeepMap’s portfolio.
On a recent episode of The Robot Report Podcast, we experienced what can happen to an autonomous vehicle when it is confused by a change in its expected route. Construction zone, accidents, and traffic are just some of the potential things that can impact an AV’s world. In a world where AV’s share a unified map and can update that map in real-time, the vehicles can prepare for and avoid issues. Work zones and road closures can be identified and then modify the world map for the benefit of all.
With the acquisition, DeepMap gains the immense resources and channel strength of Nvidia. This will likely accelerate the productization of DeepMap’s technology, but we’ll have to wait and see.
Tell Us What You Think!