Tomtom taps AI to create street map for autonomous driving

September 30, 2016 // By Christoph Hammerschmidt
Dutch navigation technology provider Tomtom will collaborate with graphics chip vendor Nvidia with the goal to leverage Artificial Intelligence (AI) to create a highly exact, cloud-based navigation platform for autonomous vehicles. Through this collaboration, Tomtom’s navigation map data will be accessible for Nvidia’s Drive PX 2 AI computing platform.

Tomtom claims to have mapped more than 200.000 kilometers (about 125.000 miles) of highways worldwide. If these data become available to car-based DrivePX 2 computers, the process or real-time localization of the vehicles and of generating map data for autonomous driving could be significantly accelerated. Along with the actual map data, the Dutch navigation expert provides RoadDNA, a set of proprietary data that reflect unique road features in a compressed way. Downloaded to the individual vehicle’s computing platform and matched with the car’s own sensor data in real time, these data make GPS data more accurate. Under the bottom line they allow lateral and longitudinal deviations of 50 cm or less, TomTom  claims.

The collaboration with Nvidia brings a powerful computer suited for this type of data analysis into the car. “Drive PX 2 gives us the possibility to provide updates to our HD map in the vehicle and in real-time.”

As a part of the agreement, Nvidia’s DriveWorks SDK integrates support to generate maps based on TomTom’s HD Map. The solution is open to third parties; carmakers and suppliers to self-driving car R&D efforts can access and use the solution.

Independent from the Nvidia collaboration agreement, TomTom announced a new service that helps drivers to identify and localize unoccupied parking spaces. On-Street Parking is planned to be implemented as a data feed and be initially available for 25 cities in Europe. It complements the company’s existing Off-stret Parking system.

For the service, the company is processing GPS data from the individual customer cars to determine where drivers are searching for parking spaces and at which time of the day. The findings are used to indicate car drivers the probability to find an on-street parking space at the time they are underway.   

Related news:

Nvidia praises AI as enabler for autonomous driving

Bosch, TomTom develop digital maps for automated cars

Delphi, Mobileye jointly develop platform for highly automated driving