[Tesla Leads, Hyundai Strikes Back] Hyundai Lags Behind but Closes the Gap With Genesis 'Level 2+' and SDV Demonstrator

Level 2+ Autonomous Driving for G90 Facelift in the Second Half of the Year
Two-Track Approach: 42dot Technology and Nvidia Collaboration

Advancing Proprietary Model 'Atria AI'
Verification Through Gwangju Autonomous Driving Demonstration

Leveraging Nvidia’s High-Performance Chips and AI
SDV Pace Car Launch Also Planned

Lagging slightly behind Tesla in the autonomous driving technology race, Hyundai Motor Group is set to mount a counterattack by applying 'Level 2+' autonomous driving technology by the end of this year. This technology, to be introduced in the upcoming facelifted model of the Genesis G90, is expected to enable hands-off driving, similar to Tesla's supervised Full Self-Driving (FSD) system.

In addition, Hyundai Motor Group plans to unveil its Pace Car, an advanced model for verifying software-defined vehicle (SDV) technologies, in the second half of the year. This is expected to signal the full-scale launch of the autonomous driving competition in the automotive market.

According to industry sources on July 15, Hyundai Motor Group is targeting the autonomous driving market as part of a "two-track strategy," focusing on both internalizing its own technologies through its software affiliate 42dot, and partnering with global tech giant Nvidia.

Hyundai Motor is currently advancing its proprietary AI-based autonomous driving model, Atria AI, which relies on camera sensors. This vision-based method, utilizing eight cameras and a front radar, is similar to Tesla's Full Self-Driving technology. From perception sensors to vehicle control, all processes are integrated and managed by artificial intelligence under a next-generation architecture that pursues an end-to-end (E2E) approach.

In particular, what sets it apart is that data collected by cameras and radar is merged into a single, massive deep learning model. Within the vehicle, a 'Neural Processing Unit (NPU)' handles the entire process from perception to decision and control as a one-stop operation.

This enables the system to acquire driving competence through data-based learning, empowering it to adapt flexibly to real-world scenarios. By leveraging a Visual-Language-Action (VLA) model, the car can independently infer road rules and surrounding contexts while driving.

For example, if it detects a school bus ahead, its sensors will recognize the yellow bus, flashing lights, and presence of children. The system then infers, "Because this is a school bus, keep a safe distance, stop and wait until the bus departs, and proceed cautiously while watching for children’s movements." As a result, the car will automatically stop or drive slowly as appropriate.

To enhance Atria AI, Hyundai Motor Group is participating in an autonomous driving demonstration in Gwangju in the second half of the year. Around 200 autonomous vehicles will be involved in this demonstration, providing an opportunity to verify Atria AI's real-world responsiveness and lay the foundation for further technical advancement.

Hyundai is also bringing in external experts such as Park Minwoo, who heads the Advanced Vehicle Platform (AVP) division at Hyundai and Kia and is also CEO of 42dot, as well as Lee Heeseok, executive director and expert in computer vision and deep learning-based perception systems at 42dot, and Jeremy Ma, managing director and head of the newly established Silicon Valley team within the AVP division. Together, they are reshaping the landscape of Hyundai's autonomous driving technologies.

These efforts are part of Hyundai's broader strategy to strengthen its external cooperative network—most notably, by forming a strategic partnership with Nvidia. Nvidia has successfully built its own autonomous driving ecosystem by providing in-vehicle computing hardware, AI/software stacks, and supporting infrastructure.

In terms of hardware, Nvidia developed the high-performance AGX Orin chip and established an integrated architecture called Hyperion by combining it with sensors and other components. Nvidia also offers the automotive operating system Drive OS, the autonomous driving software stack Drive AV, and has developed an inference AI model named Alpha-Mayo.

The company further accelerates technical advancement with DGX, a supercomputer designed for training using accumulated driving data, the Omniverse digital twin simulation platform, and the World Foundation Model Cosmos running within it—completing an ecosystem for developing autonomous driving technologies.

Jensen Huang, CEO of NVIDIA, is speaking after hearing an explanation about the mobile robot platform Mobed from Euisun Chung, Executive Chair of Hyundai Motor Group, at Hyundai Motor Group's Yangjae office in Seocho-gu, Seoul on June 8, 2026. Photo by Dongju Yoon

In March, Hyundai Motor Group signed a strategic partnership with Nvidia covering autonomous driving and future mobility areas such as software-defined vehicles (SDVs).

Under the partnership, Hyundai Motor Group will handle vehicle design and HPVC manufacturing, while utilizing Nvidia's high-performance chips and Hyperion platform for autonomous driving technology development. Hyundai will further enhance its proprietary Pleos OS and Atria AI driving model, using Nvidia's AI training infrastructure in this process. Additionally, by combining driving data collected by affiliates such as 42dot and the robotaxi company Motional with simulation data from Nvidia, they aim to accelerate data-driven learning.

Moreover, during Nvidia CEO Jensen Huang’s visit to Korea in June, the two companies also discussed expanding their collaboration beyond existing fields—including efforts towards developing the Saemangeum AI Valley, establishing physical AI, and building a global standard AI ecosystem.

Song Seonjae, a research analyst at Hana Securities, commented, "With Hyundai’s strengthening internal capabilities and its collaboration with Nvidia, the foundation has been laid to close the technology gap with Tesla and Chinese companies in the autonomous driving sector." He added, "The release of the SDV Pace Car and the 'Level 2+' G90 in the second half of the year will mark a critical point for reassessing Hyundai’s technological competitiveness based on the scope of data collection and the degree of performance enhancement through learning."

This content was produced with the assistance of AI translation services.