Hitachi: Automatic generation of maps for autonomous driving from navigation system maps

From keynote presentation at 4th ReVision Mobility Summit

2020/03/19

Summary
Issues in the commercialization of autonomous driving and ADAS functions
Hypothesis in problem-solving - 2 components for technical evolution concerned with external perception
Detailed Geometry Map (DGM) and Autonomous Driving Experience Database (ADD)
Generating the DGM from navigation system maps or high definition maps
Applicable range of ADD (Autonomous Driving experience Database)
Examples of expected effects of ADD

Summary

Improving the function and performance of autonomous driving vehicles and expanding the operating region of autonomous driving
(Source: Hitachi, Ltd., below same)

The 4th ReVision Mobility Summit was held in Tokyo in February, 2020.

This report summarizes the presentation given by Mr. Takashi Nomura, Chief Engineer, Connected Car Solutions Division, Smart Digital Solutions Business Development Division, Smart Life Business Management Division, Hitachi, Ltd. (Hitachi) on the subject “Map technology and database construction that supports the evolution of ADAS (Advanced driver-assistance systems) and autonomous driving”.

Autonomous driving systems now being developed by the major OEMs are based on high definition map information (HD-Map: High Definition Map), and run by comparing information obtained from various sensors to the high definition map information. Autonomous driving can be likened to running on a rail of the map. On the other hand, when autonomous driving is realized on ordinary roads in the future, there will be an enormous cost to expand the high definition maps which are maintained for existing highways and motorways to ordinary roads. There is also the problem that maps, once produced, quickly become out of date.

However, in the future, sensors and GNSS (Global Navigation Satellite Systems) will become more advanced and AI functionality will be enhanced with improved predictive capabilities. It is also expected that the expansion of useful content will continue; if so even if maps are not highly precise (not requiring absolute precision at the level of ± 25cm for position information, at most ± 1m) the vehicle will be able to run while utilizing nearby information obtained in real time. With this premise, Mr. Nomura of Hitachi is proceeding with a plan to generate a new concept of autonomous driving maps from vehicle navigation system maps (SD-Map: standard map), and this he calls DGM (Detailed Geometry Map) (also called high-detail map to distinguish from high definition maps).

Also, he is planning to create a database of driver experiences in areas where driving is difficult or unexpected difficulties occur to realize smart autonomous driving. This driving logic data will be called ADD (Autonomous Driving experience Database).

The concept is to expand the areas where Level 3, 4, and 5 autonomous driving can be conducted safely without limitation (i.e., on ordinary roads) by combining improved sensors and GNSS with DGM/ADD. Hitachi has already started a development project with a map maker and would like to release something in FY 2021.

This report is for paid members only. Remaining 6 chapters remaining.
Free membership registration allows you to read the rest of the article for a limited time.

Click here for a free trial.