Nvidia has agreed to buy DeepMap, a high-definition map start-up, bolstering its mapping and localization capabilities for autonomous vehicles (AVs).
To Nvidia, maps are a critical component for self-driving cars, particularly regarding safety, which will be one of the main priorities to get consumers to adopt the technology.
The acquisition of DeepMap adds AV-specific HD maps to Nvidia’s Drive platform — the company’s software-defined end-to-end AV product that includes deep neural network training and validation in the data center for compute in vehicles.
DeepMap’s technology will bolster Nvidia’s mapping and localization products so self-driving cars always know where they are and where they are going, according to Danny Shapiro, senior director of automotive at NVIDIA.
“And because DeepMap also uses crowdsourced data from vehicle sensors, we can build high-definition maps that are continuously updated as the car drives anywhere around the world,” Shapiro said. “As a result, we can help deliver software-defined AVs that get better, smarter and safer over time.”
Over-the-air updates to HD mapping allows vehicles to continuously receive new intel about the road ahead, potential hazards, traffic information and the environment around the vehicle.
Unlike humans that use navigation tools only at a few meters away, HD maps are critical to automated driving because there is a need for greater precision and accuracy. The maps must be able to operate with centimeter-level precision for accurate localization, which allows the AV to locate itself anywhere in the world, Shapiro said.
“Proper localization also requires maps that are constantly updated and reflect current road conditions such as construction or work zones, lane closures, etc., so that the AV can plan a safe path forward,” Shaprio said. “AV maps also need to be able to efficiently scale across fleets with fast processing and minimal data storage — while also being able to properly and safely function worldwide.”
The DeepMap acquisition
While Nvidia already works with several mapping companies for its Drive platform and many are in the partner ecosystem, DeepMap has developed mapping solutions that are already validated by the AV industry with a variety of potential customers worldwide.
The San Francisco Bay Area-based DeepMap has decades of collective experience in mapping technology and developed its maps for AVs for both map creators and map consumers. Terms of the acquisition were not disclosed.
Nvidia’s Drive platform will continue to be updated moving forward to include advances in computing, networking and security for AVs to bring added levels of safety and convenience.
Nvidia Drive can perform Level 2+ AI-assisted driving but will be scalable all the way to Level 5 fully autonomous vehicles, Shapiro said. Additionally, new functions and features to the platform can be developed and added via over-the-air updates.
“While it’s up to the customer to decide what they want to implement in their self-driving cars, we provide the end-to-end solutions to create different types of AVs with different intended uses, features and capabilities,” Shapiro said.