On October 16, at Hyundai Motor Group's I-Forest Center in Uiwang, South Korea, the AI agent 'Polaris' demonstrated its capabilities in automotive painting processes. When asked about the painting procedure, a virtual laboratory appeared on a large screen, showcasing the process of immersing a car body in paint.
Lee Ji-hwan, a senior official at Hyundai, explained, "We are virtually simulating the process of dipping the car body into the paint to check how uniformly the coating is applied. We are also validating the adhesion strength of the paint after application through simulations."
Polaris, developed jointly by Hyundai and Kia, is a manufacturing-specific AI agent designed to assist workers by processing requests entered through a chat interface similar to ChatGPT or via voice commands.
When asked to report on major defects in the car body and paint, Polaris generated a report detailing defect types and visual graphs, which it then emailed to the worker.
At the 'Virtual R&D Lab,' 12 virtual laboratories showcased simulations of wireless communication waves, robotic task path generation, and vehicle exterior inspection camera validation processes. This marks a shift from traditional equipment-centered production methods to a software-defined factory (SDF) that operates data-driven factories in both real and virtual spaces.
A notable example is the 'tunnel-type exterior inspection vision technology,' which checks the exterior condition of vehicles by passing them through a tunnel-shaped inspection device. This method processes approximately 5,000 images per vehicle to identify the location and size of defects.
AI learning for robots in manufacturing processes is advancing. During the demonstration, a robot learned to pick and move parts after approximately 100 repetitions, performing tasks autonomously without human intervention. Even with slight changes in part positioning, the robot maintained continuity based on its learned information.
Hyundai Motor Group is collaborating with NVIDIA to develop technology that allows robots to learn movements in virtual spaces. In a demonstration of assembling weatherstrips—rubber components that prevent noise and rainwater intrusion between car doors and bodies—an AI-trained robot used 3D vision to identify vehicle positions, attach parts, and verify quality. The speed of the robot's movements noticeably increased over time. Some production lines also showcased logistics automation technology using automated guided vehicles (AGVs).
Hyundai Motor Group aims to establish a system that verifies production processes from the new car development stage based on its AI capabilities. This includes the new Ulsan plant, which will begin operations at the end of this month, as well as domestic and international production facilities. From October 19 to 21, the 'I-Forest Tech Day 2026' will be held at the Uiwang Research Institute, showcasing over 200 related technologies.
Choi Young-tae, head of the I-Forest Center at Hyundai and Kia's Manufacturing Solutions Division, stated, "Robots will replace dangerous tasks in manufacturing, ensuring safety and achieving quality innovation through data-driven operations. The role of humans will shift from repetitive and uniform labor to areas of judgment and management."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.