Source: National Geographic Information Institute

The three German premium car companies BMW, Mercedes-Benz (Daimler), and Audi joined forces to buy a company called HERE. Hyundai Motor Company introduced a self-driving car at CES 2017 and said that this is what played the biggest role in self-driving technology. There are also talks that the Tesla accident that resulted in a fatality last year due to a self-driving car could have been prevented if 'this' had been available.

What do these companies representing the current automobile industry have in common? It is 'maps'. Here is a representative digital map company, Hyundai Motor Company's self-evaluation of its best technology is maps, and Tesla also pointed out the shortcomings of 3D maps.

Why is the need for a 'precision map' for autonomous driving being raised when cutting-edge 'sensor' technologies such as lidar and radar are rapidly developing? The industry refers to maps as 'another sensor' that can serve as a lifeline when cameras, communications, and sensors all fail.

MMS vehicle builds a map with precision (Source: National Geographic Information Institute)

There are three types of maps that go into a car. First, there are △navigation maps made with GPS and cameras based on the basic map, △ADAS maps using the Mobile Mapping System (MMS), and △high-precision road maps (or HD maps).

Navigation provides accuracy within a radius of several tens of meters (m), and ADAS maps provide information with road slope and curvature information added to this. This is a map that informs drivers about deceleration and acceleration sections. The current precision road map can recognize a radius of 25 cm and distinguish whether the vehicle is in the first or second lane.

A precision road map is created in the following order: surveying using MMS, collecting 3D data, correcting data, creating 3D LAS, mapping, editing and structuring in that order. MMS creates maps using GPS, Lidar, an inertial navigation system (INS), and a vehicle equipped with a camera. GPS receives signals at intervals of 1 to 30 seconds, and INS uses equipment with a resolution of approximately 0.05 to 0.1 degrees. INS can provide location information through acceleration when GPS is not available. The data configured in this way is processed into a map. 3D data is collected, corrected, and mapped, and a LAS map in the form of a point cloud that includes location information is created. If you create a map that includes regulatory lines, road boundaries, stop lines, tunnels, bridges, underpasses, safety markings, road markings, signals, etc., you get a high-precision road map.

National Geographic Information Institute, 471km map of 9 sections including Yeouido

Current status of high-precision road maps published by the National Geographic Information Institute

The difficulty lies in the fact that developing a high-precision road map that requires cutting-edge technology and a lot of data is not easy. The technology and budget required for map development, as well as the price of the MMS vehicle, which can cost up to 1 billion won per vehicle, are also burdensome. Accordingly, in Korea, the National Geographic Information Institute has been producing and releasing maps since 2015. Including the Yeouido and Daegu regulatory free zones released on April 30, a total of 9 sections and 471 km of high-precision road maps have been created. In addition, the National Geographic Information Institute announced that it plans to focus on building high-precision road maps for areas with national needs and demands in line with the commercialization of autonomous driving level 3 (conditional autonomous driving) by 2020 and on standardization and technology development.

Among private companies, Hyundai MNSoft is the only one that has MMS and has built a precision map. Hyundai MNSoft, which has been providing navigation Genie and location-based services, is building a high-precision map for major roads with two or more lanes across the country and is building a map in conjunction with the National Geographic Information Institute’s map for some sections.

HERE, a global company, is a leading example of precision map technology. HERE’s map consists of three layers. ▲The basic HD Map supports driving such as vehicle location information, turning radius, lane division, lane change, and overtaking with 3D road information with an accuracy of 10 to 20 cm. ▲In addition, real-time dynamic information provides traffic conditions, weather conditions, and traffic accident information in the HD Live Map and collects data by integrating it with sensor data. ▲The Humanized Driving Layer, which analyzes driver driving patterns, automatically adjusts speed on curved sections to increase safety and reliability, making it suitable for long-distance driving.

HERE supplies maps to 80% of vehicles produced in North America and Europe. At CES this year, HERE CEO Edzard Overbeek announced an agreement with NVIDIA, saying, “We plan to accelerate the development of Live Map using AI. In order to meet the most important accuracy in autonomous driving maps, up-to-date maps are needed, and Audi, BMW, and Daimler will be collecting real-time data using sensors.”

It is known that HERE is constructing maps not only in North America and Europe but also in Korea. Lee Yong-kwan, an official at the National Geographic Information Institute, said, “It has been said for a long time that HERE is constructing maps in Korea with MMS vehicles.”

Road construction, research on standards in progress

The issue of standardization for precision maps is also being highlighted. This is because maps play an important role in autonomous driving, but there is no international standard. Although some international standards are being applied, each company is creating standards, so it is a cautious move because the interests of a specific company may be reflected. In 2016, the concept standard ISO/TC 204 for LDM (Local Dynamic Map) was proposed for Intelligent Transport Systems (ITS).

“Domestic standards and technology trends are still lagging behind,” said Dr. Hyun-Sang Choi, an engineering professor at the ICT Convergence Research Center of the Korea Institute of Construction Technology. In Europe, LDM is being applied as a standard, and in Korea, road construction is conducting research on LDM at the suggestion of the ITS Association.


■Interview
Lee Yong-gwan, Office Manager / Geographic Information Division, National Geographic Information Institute

"We hope that our map service will take autonomous vehicle research to the next level."

National Geographic Information Institute, Yong-gwan Lee, administrative officer

Q. In Korea, even Google is not allowed to export maps, so can a foreign company build a domestic precision road map?
A. Security, it is necessary. Since there is no standard for precision maps yet, we had many concerns before releasing the maps. At first, we only distributed processed vector maps when we visited the information source directly and took the maps. Now, you can download them after writing down the usage on the Internet, and we also provide LAS data through agreements that pay attention to security.

Q. Does security also include privacy?
A. LAS data includes video information, but due to security issues, we are disclosing the video without it. There were some requests for this, but we provided it after deleting some sections that could be problematic.

Q. How are the companies that provided the maps using them?
A. We have received guidance from research institutes, companies, and various public institutions, but it is still in the research stage. We have started making maps to raise the level of research, so we do not regulate research. If it reaches the distribution stage, we may impose security-related sanctions. As of now, we are supplementing the map by asking about the map's shortcomings and necessary data as research is in progress.

Q. Yeouido, which was unveiled last month, is the only urban area among the nine sections. Were there any difficulties?
A. MMS vehicles receive satellite information and correct their positions, but it was difficult to determine their positions due to interference between tall buildings. To increase accuracy, the map corrects positions using survey points, and we took more survey points to accommodate the characteristics of urban areas. We also conducted research with Seoul National University, which has been studying autonomous vehicles in urban areas. Seoul National University's autonomous vehicle, Snuber, plans to operate autonomous vehicles in Yeouido.

Q. I heard the project is until 2020. Where are the newly built sections?
A. This year, it is the section from Yeongdong Expressway to Pyeongchang Olympic Stadium, Sejong City, and Pangyo Zero City. We plan to show autonomous driving as a pilot service in the Pyeongchang Winter Olympics section, so we are also planning to build a map. There are other plans, but I think we need to change them depending on the research situation. Technology development also needs to be continuous. I can't say for sure how it will be organized.

Q. What is the role of the National Geographic Information Institute in the era of autonomous vehicles?
A. Our goal is to bring autonomous vehicle research to a certain level and commercialize it. Private companies are also making maps, and there are some parts that overlap with ours. However, the maps we make are open to ‘LAS data.’