US based graphics processor manufacturer Nvidia has unveiled a new mapping platform — Drive Map — which it claims will provide mapping coverage of 500,000 kilometres of roadway in North America, Europe and Asia by 2024. The company has revealed that this map, which will be regularly updated, will be equipped with multiple localisation layers of data for use with camera, radar and lidar modalities.
Announced at a GPU technology conference (GTC) for developers, Nvidia maintained that the maps will be compatible with Level3 and Level 4 Autonomous Vehicles. The platform is based on technology developed by a startup called DeepMap, a company making detailed maps for autonomous vehicles since 2015 and which was eventually taken over by Nvidia.
Nvidia, while presenting the solution, also affirmed that the Artificial Intelligence (AI) driver can localise to each layer of the map independently, providing the diversity and redundancy required for the highest levels of autonomy. It has also clarified that the map is available to the entire autonomous vehicle industry.
Nvidia founder and CEO Jensen Huang said that Drive Map is a multimodal mapping platform designed to enable the highest levels of autonomy while improving safety. It combines the accuracy of DeepMap survey mapping (from the acquired company) with the scale of AI-based crowd-sourced mapping, as per his claims.
The camera localisation layer consists of map attributes such as lane dividers, road markings, road boundaries, traffic lights, signs and poles. Furthermore, the radar localisation layer will be utilised in poor lighting or weather, which the company claims could be challenging for cameras and lidars.
The development comes at a time when autonomous vehicles are already gaining eminence in many states of US. While Waymo (formerly the Google self-driving car project) is on course to introduce a fully autonomous-driving service in San Francisco for ferrying passengers within the city, Uber Technologies is also actively working on driverless car projects.