Discussion at Automotive Innovation Day Forum
Control Authority Confusion: No International Standards or Precedents Yet

As the commercialization of autonomous vehicles becomes more concrete, will these changes pose a threat to us or present an opportunity?

On the 5th, the 'Automotive Innovation Day 2017' held discussions on this very topic. Prior to the discussion, the event proceeded with conferences divided into three sessions: ADAS & autonomous vehicles, connected cars & interior electrical components, and autonomous driving testing & development fields. The panel discussion concluding the conference was held under the theme 'Autonomous Driving: Threat or Opportunity?' and included Professor Jung Gu-min from the Department of Electronics Engineering at Kookmin University (moderator), Jens Hinrichsen, Senior Vice President of NXP Semiconductor, Shin Jae-gon, Head of the Autonomous Driving Research Division at the Automobile Safety Research Institute, and Yu Si-bok, Center Director of the Korea Automotive Technology Institute.


Professor Jung Gu-min (Moderator) According to the report, when autonomous vehicles are commercialized, the number of vehicles will decrease. The question of whether we will move toward fully autonomous vehicles is being discussed, and it is said that autonomous driving will skip Level 3 and move directly to Level 4. What are your thoughts on this?

Director Shin Jae-gon Currently, the definition of Level 3 is ambiguous from a technical standpoint. The NHTSA levels are based on general concepts, so there are many ambiguous aspects technically. The fact that OEMs are developing without a clear definition of Level 3 ultimately leads to issues of control authority responsibility. The key question is 'at what point is control transferred?' The timing and method of transferring control lack clear definition in international standards. These aspects must be clarified before we can say we are moving to Level 4.

Director Yu Si-bok In the ISO standardization meeting held in Paris in April, the issue was raised: if the system is monitoring a driver who is doing something else and the situation requires transferring control authority, what happens? It was suggested that if we give several warnings and then transfer control, the vehicle should not bear responsibility. Since there are no precedents related to control authority yet, this requires discussion.

From Level 4 onwards, the vehicle must bear responsibility. In Japan, if an accident occurs and a ruling is made that the vehicle is at fault, under current law individuals must prove the vehicle's fault. This would mean compensation from accidents exceeds revenue from vehicle sales, which would prevent market formation.

Professor Jung Gu-min (Moderator) In Level 1 and 2 autonomous vehicle accidents, driver negligence was the cause. Drivers bore legal responsibility. But as we move to Levels 3 and 4, laws will need to change as well. What is the roadmap?

Director Shin Jae-gon In developing Level 3 technology, developers can say it is Level 2. If NHTSA registers it as Level 2, the driver must bear responsibility. When transferring control to the driver, they must always be attentive—these aspects need to be clarified along with technology development.

Trend of Providing Software and Services Together

Professor Jung Gu-min (Moderator) When autonomous vehicles are commercialized, which companies and businesses do you think will benefit the most?

Director Shin Jae-gon Autonomous vehicles and electric vehicles have the same ultimate goals, but autonomous vehicles are more profitable than electric vehicles. Electric vehicles naturally form markets as charging infrastructure develops, but autonomous vehicles follow a flow where ADAS systems, sensors, and artificial intelligence come together, making the market broader. In the short term, software companies will profit. In the long term, suppliers that provide both software and services will profit. Looking at smartphones, I believe ultimately service providers will be positioned above OEMs.

Director Yu Si-bok According to 2015 UK research, by 2030, autonomous vehicles will generate a ripple effect of 70 trillion won. Over 30 trillion of this is because people can do other things inside the vehicle due to autonomous driving. I believe companies providing these services will lead the market.

Professor Jung Gu-min (Moderator) Currently, most of Korea's automotive business is component business. What preparations should component suppliers make?
Director Shin Jae-gon Component suppliers need to think like OEMs or Tier 1 suppliers. They must develop software and go all the way to the OS level. Hyundai Motor also has to use another company's OS, so it becomes dependent. Looking broadly, I think they should respond by separately developing OS systems.

Professor Jung Gu-min (Moderator) From the perspective of finished vehicle manufacturers, is autonomous driving a threat or an opportunity?

Director Yu Si-bok The autonomous vehicle industry itself is definitely a major threat to the automotive industry. To become an opportunity, we need to improve our structure. In the case of vision sensors, 80% are imported. For an industry that is gradually growing, this is unsuitable. However, it will become an opportunity only if we nurture rather than abandon the component industry.

"It will become an opportunity only if we nurture rather than abandon the component industry"

Professor Jung Gu-min (Moderator) Do you think the commercialization of fully autonomous vehicles is feasible?


Jens Hinrichsen It will eventually be implemented. The entire ecosystem—components, hardware, software—is completely changing. Through collaboration, we will advance the technology to Level 5. For commercialization, many customers need to want and accept it. Therefore, reliability is important. Legislation and systems that clearly determine responsibility must be established. Ethics, law, regulation, and technology must move forward together for commercialization to be possible.

Professor Jung Gu-min (Moderator) If there is a core technology for commercialization, what would it be?

Jens Hinrichsen We need various sensors that will form the basis of vision, and we need WiFi and 5G technologies. Next, we need safe, secure, and highest-quality functional safety for autonomous driving to be achieved.

Professor Jung Gu-min (Moderator) What activities are being conducted in Europe for commercialization?

Jens Hinrichsen Currently, there are many regulatory-related activities. Germany's Minister of Transportation released autonomous vehicle guidelines. Many technology collaborations are also taking place. We are working with partners on radar, software, security, V2X, and more.

Director Yu Si-bok What activities is NXP engaged in regarding V2X?

Jens Hinrichsen Based on NXP's WiFi solutions, we provide strong regulatory support. We also see 5G as usable in a supplementary capacity. V2X testing has also been conducted on the German Autobahn.

Professor Jung Gu-min (Moderator) Technology and systems must move forward together. How are legal systems and guidelines for each level being prepared?

Director Shin Jae-gon Step-by-step legislation is difficult for Korea alone. International discussions are necessary globally. Rather than step-by-step, UN regulations are defining technical definitions and standards. OEMs want the government to provide guidelines. Currently, autonomous vehicle technology is advancing too rapidly for lawmaking. We are discussing how not to let laws become a bottleneck when commercialization arrives. If developers understand the direction that international regulations are moving and increase reliability accordingly, that should work.

K-City (Simulated City) for High-Speed Driving Tests to Open This Year

Professor Jung Gu-min (Moderator) There are high expectations for K-City. How will it be utilized?

Director Shin Jae-gon When developing systems or vehicles equipped with autonomous driving, evaluation and validation are absolutely necessary. Since evaluation and validation are difficult on regular roads, the Ministry of Land, Infrastructure and Transport is building an autonomous vehicle experimental city within the Automobile Safety Research Institute's 110,000-pyeong site. Configured as a sort of virtual city, it enables all simulations including highways, rural roads, urban streets, school zones, and construction zones. Plans are to have dummies and robots simulate pedestrians and vehicles for repetitive experiments and various scenarios. Communication networks will also be installed up to 5G, allowing high-speed communication performance testing. This year, only the highway section will open, with the remainder planned to open around the end of next year.

Floor Question What are the usage conditions or pricing for K-City?

Director Shin Jae-gon We are still considering when to open it. The virtual city itself has levels. In its current state, only high-speed driving will be completed this year. Specific plans on how to actually utilize it have not yet been finalized. The virtual city's scale must be accurate. It will be upgraded in stages, but we are unsure how much will be completed by year-end.

Question Regarding driver biometric signals inside vehicles, what is the current progress and how do you think it will be applied to automobiles in relation to law?

Director Yu Si-bok There are various methods for measuring biometric signals. Brain waves are measured, and recently I have seen blood flow measured through the nose bridge of glasses. Facial temperature distribution and muscle shape are also observed via camera, and there are methods involving attachment to the steering wheel. Regardless of the method, measurement is difficult depending on each person's condition. Overall, much time will be needed for commercialization. I think voice analysis and conversation analysis will likely be commercialized first.

Director Shin Jae-gon The development phase and the actual application phase in vehicles are completely different. We are developing eye trackers to prevent drowsy driving, but actual implementation is difficult. The system is good, but ultimately what comes after drowsy driving? Lane departure or collision. When inter-vehicle distance becomes shorter, control authority should transfer, but we need to examine whether, as the eye tracker sees it, the driver is keeping eyes open while doing something else or looking forward properly. For practical installation, both reliability and necessity must be established.

Professor Jung Gu-min (Moderator) As vehicles reach Level 3 and 4, the vehicle bears responsibility. For this to happen, driver monitoring is essential. There are discussions about mandatory installation of monitoring and the need to change systems.