[Interview] Professor Jungkyu Park: "Hyundai's Autonomous Driving Will Succeed Even by Closely Following Tesla"

Interview with KAIST Graduate School of Technology Management Professor
Active Adoption of Tesla's E2E Approach Is a Practical Alternative
China's 'Swarm Strategy' Accelerates Industry Learning
Discord Between HW and SW Camps Is Normal Duri

"It would be a success for Hyundai Motor just to closely follow Tesla in autonomous driving. The advantage of being a latecomer is that you can take shortcuts. We need to adopt the Tesla approach, namely End-to-End (E2E), as our model."

Professor Jungkyu Park, concurrently serving at the Graduate School of Technology Management at KAIST, emphasized this point during an interview with The Asia Business Daily on January 12. Professor Park is an expert with extensive experience in both automotive and manufacturing fields at Hyundai Motor and LG Electronics research labs. He has analyzed industry changes from the perspective of shifts in product architecture and organizational and development methods, rather than simply technological competition. Recently, he published a book titled "The Smart Car Hegemony War," in which he examined how Software-Defined Vehicles (SDVs) and Artificial Intelligence (AI) are reshaping the competitive landscape of future mobility.

Professor Jungkyu Park of KAIST Graduate School of Technology Management is being interviewed on the 12th at Hyundai Motorstudio Seoul in Gangnam-gu, Seoul. Photo by Jinhyung Kang

The Advantage of Latecomers: Taking Shortcuts

Professor Park diagnosed that the rapid progress of AI has brought the automotive industry to a major turning point. He stressed that it is now time for Korea’s autonomous driving industry to actively embrace the Tesla-style E2E approach. "It may be a realistic choice to quickly follow Tesla’s method, which has already gone through trial and error," he said. "The very fact that Hyundai Motor switched to the E2E approach around 2023 is significant in terms of direction."

Currently, autonomous driving technology is largely divided into rule-based and E2E approaches. The rule-based method involves humans meticulously defining driving rules, which provides high safety but has limitations such as high dependence on expensive sensors and increased system complexity. In contrast, the E2E approach allows AI to make decisions by learning from data. Although it is difficult to explain the process, it excels at adapting to complex real-world road environments. Professor Park pointed out, "The stage where increasing the number of sensors leads to the completion of autonomous driving has already passed. Now, the focus should be on enhancing the 'intelligence' of vehicles."

Professor Park cited the rapid growth of Chinese autonomous driving companies as the basis for a strategy to catch up with Tesla. "Chinese companies such as Nio, Li Auto, and Momenta, led by IT entrepreneurs, have quickly imitated the Tesla approach and secured competitiveness," he explained, describing this as China’s "Swarm Strategy" for smart car technology growth.

This concept involves many companies simultaneously conducting experiments and challenges, and then quickly imitating and spreading proven methods, thereby accelerating the learning speed of the entire industry. Professor Park stated, "Part of the reason we failed to fully recognize China’s rise was because we underestimated them. Now is the time to analyze objectively and identify what we need to learn."

Professor Jungkyu Park, concurrently serving at the Graduate School of Technology Management at KAIST, is being interviewed on the 12th at Hyundai Motorstudio Seoul in Gangnam-gu, Seoul. Photo by Jinhyung Kang

Discord During the Autonomous Driving Transition Is Normal

Regarding the clash of worldviews between software (SW) and hardware (HW), and between computer engineering and mechanical engineering in traditional automakers, Professor Park offered a positive assessment. "It is natural for discord to occur as automakers with a strong mechanical engineering foundation transition to SDVs," he said. "There is no need to view it too negatively." He added, "Once products start reaching the market, points of coordination will emerge, and consensus will be found."

Professor Park explained that automotive development in the SDV era involves building up several layers of technology. These include the ground layer responsible for powertrains such as electric and hybrid vehicles, the upper air layer operating in the cloud and virtual space, and the lower air layer that connects the two. In particular, the lower air layer manages the vehicle’s operating system (OS) and interfaces, serving as a bridge between the physical and virtual worlds. Recently, AI has been added as an upper layer, learning from data at each layer to enhance the vehicle’s decision-making and functionality.

Professor Park observed that while smart car companies in the US and China have achieved results through top-down development led by IT entrepreneurs, Korea must build a structure that rises from a manufacturing base on the ground to the upper layers, inevitably creating a significant gap between the two approaches. He said, "If Hyundai Motor, as a legacy company, can establish such an integrated structure, it will become a new model for innovation."

He pointed out a structural limitation in Korea’s autonomous driving industry: the tendency to regard SW talent as external personnel. Professor Park explained that the gap in SW industry competitiveness between the United States and Japan also stems from whether SW talent is internalized. According to the "DX White Paper 2023" by Japan’s Information-technology Promotion Agency (IPA), the proportion of SW talent employed as regular internal staff is 65% in the US, but only 26% in Japan.

Professor Park emphasized, "Without embracing SW talent within companies, it is difficult to expect leading innovation in SDVs and autonomous driving. It is also important to note that many leading US and Chinese companies in the SDV field have grown under the leadership of individuals with no prior automotive experience."