NVIDIA Targets Gwangju Robotaxi Pilot, Using South Korea as a Springboard

NVIDIA is seeking to join South Korea’s government-led driverless robotaxi demonstration program, with Gwangju designated as its first test area. According to Sina Finance, the company has approached several South Korean autonomous-driving firms and presented a proposal to the Ministry of Land, Infrastructure and Transport. The proposal would introduce NVIDIA’s Hyperion autonomous-driving platform and HaloS safety system into the ministry’s robotaxi pilot. Hyperion combines an onboard AI computer, cameras, radar and other sensors, and supporting software. HaloS is designed to address system failures and risky scenarios and to validate autonomous-driving safety. If the proposal is adopted in Gwangju Comprehensive Special City, South Jeolla Province, autonomous vehicles Hyundai has built on the Ioniq 5 platform could use NVIDIA’s technology. RideFlux and Autonomous A2Z (A2Z), selected for the Gwangju pilot in April, are expected to integrate their own autonomous-driving software with NVIDIA’s platform. KAR, TransTech and Mobiltech, which provide vehicle conversions, sensors and data services to the three companies, could handle related work, including installing NVIDIA’s platform, connecting it to existing systems and integrating sensor data. NVIDIA entered the autonomous-driving market in 2015 with its Drive PX onboard AI computing system. It has since expanded into road-scene AI models, simulation and validation tools, and integrated computing and sensor platforms, aiming to establish its technology as a core platform for automakers and autonomous-driving developers. This year, NVIDIA released its Alpamayo autonomous-driving AI model in January and announced a plan with Uber in March to expand driverless robotaxi services to 28 cities worldwide by 2028, starting in Los Angeles and San Francisco. The companies plan to collect road-condition data and train and validate AI models. South Korea’s Ministry of Land, Infrastructure and Transport plans to add four to five pilot cities in 2027 after Gwangju, expanding the vehicle fleet from 200 to 3,000. NVIDIA would have opportunities to test its AI models and onboard computing platform on roads in multiple South Korean cities. Some local companies are weighing the value of cooperation because they have developed their own sensors, computing hardware and safety systems. One industry insider said adapting existing software to NVIDIA’s platform would be necessary, while the benefits remained unclear. Other firms said they were positively evaluating cooperation, citing the potential for NVIDIA to become a mainstream technology standard in the global autonomous-driving ecosystem.