Tesla’s self-driving taxi shocks industry, sparks reflection in Korea’s R&D sector

Token development projects should be abandoned. Korea must overhaul outdated regulations and devise a bold national roadmap for the future of autonomous technology.

Seo Seung-wooThe author is a professor of electrical and computer engineering at Seoul National University.

Tesla’s long-promised robotaxi service officially launched in late June in Texas. After several delays and missed deadlines in past years, the company finally delivered, meeting long-held market expectations. What sets this debut apart is not simply that Tesla rolled out autonomous vehicles — it’s that it did so using only inexpensive camera sensors combined with artificial intelligence. In terms of commercial viability and scalability, Tesla may be the first to have successfully achieved this.

The impact on the automotive industry could be profound. Tesla’s self-driving software, integrated into its already dominant electric vehicle lineup, opens new avenues for both private use and commercial mobility. Owners of Tesla vehicles may soon be able to earn income by letting their cars operate like Uber taxis, without a driver. This not only increases the appeal of Tesla vehicles but also positions the company as a potential platform provider connecting drivers and passengers.

A Tesla robotaxi drives on the street along South Congress Avenue in Austin, Texas, on June 22. [REUTERS/YONHAP]

I recently had the opportunity to test Tesla’s autonomous driving capabilities in San Francisco, which included navigating both city streets and highways. The experience was, in a word, shocking. While Tesla classifies its self-driving system as Level 2 — requiring driver supervision — what I encountered felt closer to Level 4 autonomy.

Once I entered the destination, the car began driving entirely on its own. There was no need to steer, accelerate or brake, as long as I kept my eyes forward. The car changed lanes to avoid slow traffic, merged onto highway ramps, recognized blinking signals, yielded at intersections and parked in tight alleyways — all without input. Within ten minutes, the initial unease gave way to surprising comfort. It reminded me of my experience with Google’s Waymo robotaxi just a year earlier.

With this level of functionality, navigating unfamiliar roads or worrying about local traffic rules could soon become obsolete. Though current regulations still require drivers to remain attentive in case of accidents, from a technological perspective, Tesla’s autonomous cars already seem fit for everyday use.

Tesla’s breakthrough did not come overnight. Its success lies in a feedback loop: cameras installed in sold vehicles collect vast amounts of real-world data, which improves driving algorithms. These updates are then sent wirelessly to all vehicles. A decade ago, few believed that full autonomy could be achieved using cameras alone. Tesla CEO Elon Musk and his engineers were ridiculed for attempting what many called a reckless gamble. In hindsight, their vision and technical prowess now command admiration.

For more than 20 years, many researchers — myself included — believed that full autonomy required a combination of sensors: cameras, LiDAR, and high-definition maps. We thought these systems had to complement one another. Tesla’s recent performance shattered that assumption. It made me reflect on the limits of our thinking and my own biases. I had underestimated the potential of alternative approaches and neglected the importance of embracing imperfect but promising technologies in early stages of development.

A Waymo autonomous self-driving Jaguar electric vehicle sits parked on the side of the road after the path of travel was blocked by fire hoses and emergency vehicles as Los Angeles Fire Department (LAFD) firefighters responded to a fire in a dumpster in Los Angeles, California on May 8. [AFP/YONHAP]

Autonomous driving has moved beyond slow incremental progress and is approaching a tipping point. Korea now faces two choices. One is to adopt proven foreign technologies and install them in domestic vehicles. The other is to follow Tesla’s path, building similar systems from the ground up. The second option is far more difficult, as it requires replicating Tesla’s hard-earned know-how, accumulated over years of iteration and data collection. Either way, Korea must commit sustained, large-scale investment in time and capital — or risk becoming an observer in a global shift.

Over the past seven years, the Korean government has poured more than 1 trillion won (around $726 million) into autonomous driving research and development. That investment served as a primer for ecosystem building. But as we move toward commercialization, a clear strategy is still missing. Token development projects should be abandoned. Korea must overhaul outdated regulations and devise a bold national roadmap for the future of autonomous technology.

Translated from the JoongAng Ilbo using generative AI and edited by Korea JoongAng Daily staff.