Exploring the Future of Autonomous Driving: A Forbes Interview with HanBin Lee on Company Highlights and Vision
Seoul Robotics is the sole start-up among global autonomous driving software companies that has successfully commercialized the products. Led by CEO Hanbin Lee, the company excels in infrastructure-based autonomous driving and is now aiming at the global automotive logistics market, which worth tens of trillions of won.
November 23rd, 2023
In November 2019, the early winter air in Helsinki, Finland, was brisk as the SLUSH event took place, a global startup festival held annually in Finland. Despite the chilly weather, the event's venue buzzed with energy from over 10,000 technology-based startups from around the world, attended by venture capitalists, major corporations, research institutions, and government officials from various countries, all eager for innovation.
Hanbin Lee, CEO of Seoul Robotics, was among the attendees, filled with ambition to advance the era of autonomous driving. His company, which specializes in AI-based autonomous driving software, was confident in their expertise. This confidence and sincerity caught the attention of BMW, a global automotive manufacturer, which approached them with a proposal for developing an autonomous driving system capable of operating perfectly in confined spaces like factories and parking lots, even under adverse weather conditions such as snow or rain.
BMW's proposal was clear and straightforward, asking Seoul Robotics to create a system that could achieve flawless autonomous navigation in specified environments. Lee's affirmative response, backed by his confidence in his company's LiDAR-based autonomous driving software, led to a partnership with BMW. This was especially notable because BMW had previously approached a global tech company in their home country with the same proposal, only to be told it wouldn't be feasible until 2025. Lee's eager and prompt reply, "We'll try it by next week," was a pivotal moment that solidified the collaboration between Seoul Robotics and BMW.
Pivot to Infrastructure-based Autonomous Driving
"Of course, we were delighted and thankful for BMW's proposal. However, it was quite surprising as it was beyond our expectations and differed from the business direction we had been dedicated to since our founding. At that time, we focused on vehicle-based software for autonomous driving, like Waymo or Tesla. But BMW's suggestion was a completely different concept—it was based on 'infrastructure' for autonomous driving. We started it thinking, 'It might just be a side project, but let's try it,' and now it has become the main business of Seoul Robotics."
"The autonomous driving market is a battleground where global tech companies and automotive manufacturers fiercely compete and sometimes collaborate to achieve the dream of 'Level 5' autonomy, which requires no driver involvement. Most of these efforts are focused on vehicle-based autonomous driving, where each car is an individual platform that operates autonomously amidst countless vehicles, people, and various obstacles on public roads. This method requires equipping every autonomous vehicle with sensors like LiDAR and operational software."
"Since its inception in 2017, Seoul Robotics has concentrated on developing vehicle-based autonomous driving software, driven by the hope and expectation that the 'Level 5' era would soon begin. The arrival of the autonomous driving era was not just a dream for CEO Lee. Everyone cheered at the sight of drivers taking their hands off the steering wheel on highways or taxis running without drivers. However, realizing this dream has taken much longer than expected. CEO Lee confessed, 'I increasingly felt that it was an insurmountable barrier.'"
"Even Tesla's 'Autopilot,' reputed to be the most advanced in autonomous driving technology, is actually only at Level 2. There have been reports of achieving Level 3, but these are mostly just for testing. The more we engaged in this business, the more we realized how far away perfect autonomous driving at Level 5 is."
This year, San Francisco became the first city in the world to authorize autonomous taxi operations, revealing the ongoing challenges of achieving Level 5 autonomy. This pilot, showcased by Google's Waymo and GM's Cruise, became notorious just weeks after launch, causing disruptions in the city and even accidents with emergency vehicles. Despite GM's claims of a flawless record over 5 million kilometers, the increase in 911 calls due to the taxis’ erratic behavior highlighted significant setbacks. CEO Lee had been as optimistic about a rosy future for autonomous driving as any engineer.
Lee believed technology would soon enable perfect autonomous driving, but realized the market was opening too slowly, thinking, "It's still far from making money with this." Amid deepening concerns, Lee received a proposal from BMW in 2021, pivoting entirely from vehicle-based to infrastructure-based autonomous driving software, changing the business focus naturally.
Infrastructure-based autonomous driving involves using infrastructure elements like poles, streetlights, or ceilings for sensor placement instead of vehicles. This system detects vehicles in confined areas like parking lots, with central software sending driving commands. The advent of electric vehicles has accelerated this shift, enabling vehicles to operate as drones, controlled directly by a central command through infrastructure-placed sensors, all managed by Seoul Robotics' AI-based software.
"When you press the accelerator in an electric vehicle, it is immediately controlled by a computer. The era of opening and closing throttle valves is coming to an end. In infrastructure-based autonomous driving, each vehicle operates like a drone receiving commands from a central authority. The vehicle recognition by sensors installed in the infrastructure, the decision-making for driving, and the control of various variables are all managed at once by Seoul Robotics' AI-based software."
The only global competitiveness recognized by BMW
"There are countless companies related to autonomous driving. However, according to CEO Lee, companies that operate software for infrastructure-based autonomous driving are rare even on the global stage. Especially outside narrow interiors and in large outdoor spaces, Seoul Robotics is the only one in the world equipped with infrastructure autonomous driving technology. Moreover, they are the only company to have generated revenue and commercialized this technology.
"We've completely abandoned the pursuit of autonomous driving on public roads. We realized the variables involved are infinite. For example, in Australia, you have to respond to kangaroos, and near baseball stadiums, to flying balls. On the other hand, in car factories or parking lots, you don't have to worry about a three-year-old child crawling under the cars. Public road autonomous driving seems far off, but using infrastructure for autonomous driving in a confined space is sufficient with current technology."
CEO Lee chuckled, "If it weren't for BMW's counterproposal that came like fate, we would still be chasing the dream of public road driving." However, turning crisis and despair into opportunity was a result of recognizing the technological competitiveness unique to Seoul Robotics.
Seoul Robotics introduced the 'SENSR', a 3D perception platform after its establishment in 2017, which realized computer vision based on 3D for the first time. It uses AI deep learning to analyze 3D information obtained from sensors like LiDAR, capable of tracking and identifying moving objects within a 4cm range in a parking lot bustling with hundreds to thousands of cars. Especially, it has succeeded in implementing precision that only recognizes the movement of targets even in harsh weather conditions like heavy snow or rainstorms, thanks to the world's most advanced weather filtering AI.
"Perceive, judge, and control. These three are the fundamentals of all autonomous driving technologies, with perception being the most crucial. You need to accurately determine where objects are and how they are moving, but in software, heavy snow acts as noise that obstructs perception. Our system, which functions uniquely well in bad weather day or night, has proven its scalability at factories or parking lots for BMW."
Currently, Seoul Robotics' ATI (Autonomy Through Infrastructure) solution, 'LV5 CTRL TWR', is installed at the BMW i7 and 7 Series production plant in Dingolfing, Germany. It's the world's first autonomous driving system installed inside a car factory. About 200 sensors installed within the factory simultaneously monitor complex vehicle movements, centrally controlled by AI software. There is no need to equip individual vehicles with sensors. The infrastructure-based software turns non-autonomous vehicles into autonomous ones.
"A 1 km² area is sufficient with 200 LiDAR sensors. Autonomous driving for 300,000 vehicles produced annually is perfectly achieved within the factory. Typically, between 1,000 to 2,000 vehicles move, and autonomous driving is carried out by our system from the factory to the storage yard and up to the point of transfer to trains. With 4cm level perception, parking is even more accurate than done by humans."
CEO Lee's explanation hints at why BMW was keen to partner with a Korean autonomous driving startup first. Primarily, it was to cut costs. Previously, humans were responsible for moving vehicles produced in the factory to storage yards and then onto ships or trains. The annual driver labor costs alone for a factory producing 300,000 vehicles amount to about 20 billion won. Naturally, accidents are bound to happen when tasks are performed by humans.
"Give us your infrastructure generously for technology validation (PoC). You can do the publicity as you like, and keep all the technology IP. Just complete the autonomous driving within the factory. That was BMW's proposal. After completing the system, we didn't make a lot of money from it (laughs). However, having completed the world's largest infrastructure-based autonomous driving system, especially applied to BMW, has been a tremendous reference for us."
Vehicles produced at the BMW factory in Munich, Germany, operating unmanned using Seoul Robotics' autonomous driving technology.
Aiming at the automobile and truck logistics market
Seoul Robotics is the only company in the world that has successfully commercialized infrastructure-based autonomous driving software. The joint venture began in 2017 when CEO Lee and three other co-founders came together during an autonomous driving coding competition. It all started with an AI study group Lee founded on Facebook right after his military discharge in 2016.
"Ever since I was a child, I dreamed of making robots. After graduating from elementary school, I went to study in the United States on my own, majoring in mechanical engineering at Penn State. But once I got there, I realized that software was more crucial in making robots move. Most overseas robotics companies are also software companies. It wasn't until I was about to graduate, lamenting that I should have gone into computer engineering, that I started studying AI software in the military."
Lee said that working at the Ministry of National Defense was a pivotal opportunity that shifted his focus to software, thanks to the online courses on AI and software he devoured at the cyber cafe within his unit. After graduation, he formed a Facebook study group on AI deep learning and autonomous driving, and those he met there are still with him at Seoul Robotics today.
"In 2017, a major competition focused on autonomous driving perception called 'Self-Driving Car Challenge' was held in Silicon Valley. They provided three types of sensors—LiDAR, cameras, and radar—but we used only LiDAR and ranked 10th among 2000 teams. This convinced us that we could excel in this field, and the four members who participated in the competition decided to start a company. We thought, 'If we can place 10th with just LiDAR, what if we expanded it with software? Could we be number one?' That belief has brought us this far."
Lee's prediction was spot on. Their 3D autonomous driving model using deep learning AI boasted the highest accuracy in the world. With the advent of the electric car era and ICT companies and automakers jumping into the market, he anticipated the market would soon open up significantly. However, that was just an expectation.
"We had to make money since we started a company. We couldn't just rely on investments forever. The encounter with BMW led us to pivot to infrastructure-based autonomous driving, and in retrospect, it was the best decision."
On average, a car goes through about five parking lots from production to delivery to the consumer. Even limiting it to domestic sales in South Korea, it reaches three to four lots. Lee explained that Seoul Robotics' autonomous driving system could be applied in all these stages. It didn't take long for the strategy of completely shifting their business target to take effect, a stark contrast to the endless wait for the vehicle-based autonomous driving market to open.
"Since 2019, we have worked exclusively with BMW for three years. Even now, hundreds of vehicles are autonomously moving daily in the German factory, and there has not been a single accident during this period. Recently, we completed discussions with major automobile manufacturers in Korea and Japan. After gaining recognition in Europe (BMW), the Asian market is opening up quickly."
Lee acknowledged that "our business scope has indeed narrowed compared to our initial goal of targeting autonomous vehicles worldwide," yet emphasized that Seoul Robotics is the only place that has succeeded in autonomously driving hundreds to thousands of vehicles 24/7 within a defined space. The growth potential of the infrastructure-based autonomous driving market is also seen as an enormous opportunity. According to Seoul Robotics' estimates, the global automotive logistics market is worth about 20 trillion won, and the truck logistics market, including companies like Amazon and Coupang, is approximately 90 trillion won. These are all targets precisely aimed by Seoul Robotics.
"There are about 5,000 automotive OEM logistics hubs worldwide, and just in the U.S., there are about 400,000 truck logistics hubs. Adding Europe and Asia, it's a huge market. Autonomous valet parking is also possible in indoor parking lots. Autonomous driving will take place even within electric vehicle charging areas. Just one system in a car factory is typically worth about 100 billion won. We expect that operating our currently secured pipeline alone could generate annual revenues in the hundreds of billions of won."
Autonomous Driving Setting Forth the Smart City
One of the four co-founders of Seoul Robotics is a foreigner. They first met online (via Facebook) and met in person for the first time at a Silicon Valley competition in 2017. CEO Lee also had all his foundational life experiences in the United States from middle school until just before his military enlistment. As over 40% of the employees, who are mostly engineers (developers), are foreigners, the company’s official language and documents are all in English. One might wonder why they chose to name the company and its mission after 'Seoul.'
"We wanted to show that a world-class software company could emerge from Korea. In fact, in the global market, if communication, namely English, is perfect, the country of origin isn't really important. Look at the companies that sustain Korea; they are all based on technology. Samsung, Hyundai, SK are all examples. Yet, most Korean startups, even the unicorns, are primarily local-based. From the beginning, we built references in the global market and based on that, we approached the domestic market. If 50-60 years ago, Germany sent miners and nurses, now we are sending software."
The infrastructure autonomous driving system proven in Germany has expanded to South Korea and Japan this year. CEO Lee ultimately aims at the US and China, the world's largest automotive markets. Especially China, which has become the largest car manufacturing country in the world with the advent of the electric car era.
Starting with BMW, they are accelerating their approach to the European market. For this, they are discussing the acquisition of a Swiss autonomous driving software startup. It is a company with strengths in autonomous driving control technology. This first merger and acquisition (M&A) after founding, choosing a European company, has a clear intention. They plan to strengthen their sales in the European market using a company familiar with European culture, market, and business environment. In the Asian automotive market, excluding China, Korea and Japan are overwhelmingly dominant. CEO Lee revealed a blueprint saying, "After targeting all of Asia and Europe, we will face the US and China head-on."
Seoul Robotics is a rare example of a domestic startup where about 40% of the staff are foreigners. This is also a key factor behind their successful entry into the global market.
Fundraising for business expansion is also progressing smoothly. According to CEO Lee, Seoul Robotics has attracted a cumulative investment of about 40 billion won since its inception, with about 30 billion won gathered during the last year's investment winter. Seoul Robotics, planning an IPO in 2025, intends to raise an additional 60 to 80 billion won in a pre-IPO round next year.
"Initially, we aimed for a NASDAQ listing, but now we are looking at KOSDAQ. Considering there are hardly any software-based startups among autonomous driving companies in Korea, we thought it would be meaningful to list on the Korean market. We also considered the reality that the Korean market is one of the few places as friendly to robotics companies. The funds raised through the pre-IPO and listing will be used to enter the US and Chinese markets. As labor costs in China are rising significantly, if they start losing significantly to labor markets in India or Vietnam, we see it as an opportunity opening up quickly for us."
What is the ultimate goal of the head of one of the world’s most skilled autonomous driving software companies? CEO Lee said, "I still dream of autonomous vehicles cruising on public roads." However, he plans to achieve this with the current infrastructure-based software. In fact, a major Japanese car company recently proposed collaboration to develop a city-wide control system.
The concept of infrastructure-based public road autonomous driving systems is ongoing. In Chattanooga, Tennessee, more than 100 smart intersections equipped with Seoul Robotics technology are planned to be established. These intersections, introduced for pedestrian safety, will be the largest smart intersection network in the world, not just in the U.S. They are the first smart intersections designed to warn against wrong-way drivers, detect and respond immediately, and dramatically reduce pedestrian casualties.
"The software system that achieves perfect autonomous driving in a confined space is unique to Seoul Robotics in the world. Competitors may emerge, but we will maintain our unique leading position through market preemption and technology advancement. Just as Windows is the dominant player in the computer operating system (OS) market, the day is not far off when Korean-made software will dominate the global autonomous driving market."
The content above is the translated version of Interview with Forbes by Reporter Jinwon Jang (장진원, jang.jinwon@joongang.co.kr). For original content, please click on the link below.
https://jmagazine.joins.com/forbes/view/338785
