KG Mobility has signed a Letter of Intent with the Electronics and Telecommunications Research Institute (ETRI) and Sodis Co. to develop next-generation end-to-end (E2E) autonomous driving artificial intelligence (AI) technology.
The signing ceremony took place on the 16th at ETRI's Daejeon campus. Attendees included Kwon Yong-il, head of development and production at KG Mobility; Choi Jeong-dan, director of ETRI; Sodis CEO Kang Chan-ho; and other officials. The parties agreed to actively cooperate to develop and commercialize next-generation autonomous driving AI.
Under the partnership, KG Mobility will provide its vehicle autonomous-driving system, real-world driving data and test infrastructure. ETRI and autonomous-driving specialist Sodis will leverage those assets to build a general Driving Intelligence — an AI that perceives the environment in a human-like way and directly controls the vehicle.
Traditional autonomous driving stacks separate perception, decision-making and control across distinct subsystems. This project targets an end-to-end architecture in which a single integrated AI model interprets road conditions and simultaneously determines steering, throttle and braking commands.
The team plans to apply multimodal techniques — a leading trend in global AI — so the system combines camera-based visual inputs with language-like concepts and contextual reasoning to support more nuanced decision-making.
Key initiatives include developing reinforcement-learning-based autonomy software that predicts hazards in congested traffic and selects optimal maneuvers, and creating core perception and decision AI software for operating in hard-to-predict conditions such as severe weather.
ETRI brings foundational technologies developed through national R&D programs with the Ministry of Science and ICT, the Institute for Information & Communications Technology Promotion and the Autonomous Driving Technology Development Innovation Project Group. KG Mobility expects to fuse those technologies with its real-world vehicle data to develop optimized autonomous driving models.
“End-to-end” refers to managing or testing an entire process — from product planning and production through delivery to the end customer, or from software input to output — as a single integrated system.
A KG Mobility spokesperson said, “Our technical cooperation with ETRI will allow us to integrate advanced autonomous functions more efficiently from the vehicle development stage, helping overcome current limitations and improve product competitiveness. We will continue to advance autonomous driving technologies and complete our eco-friendly vehicle lineup to secure future growth engines and establish a sustainable growth foundation.”
Reporter Cheon Won-gi 1000@viva100.com
