Ivex Selected for Ministry of SMEs and Startups Smart Factory Project... Applies Physical AI to Husteel Precision
TECHWORLD ·
✦ AI Summary
Ivex announced on the 5th that it had been selected as a supplier for the Ministry of SMEs and Startups' "2026 R&D Performance Dissemination Smart Factory Build Support Project."
Ivex will build an AI-based inspection system and a logistics automation system at Husteel Precision's hot-forging scroll production site.
The target application is a hot-forging scroll for automotive compressors, and the company plans to expand its scope from there to complex 3D-shaped parts and misaligned parts.
Ivex announced on the 5th that it had been selected as a supplier for the Ministry of SMEs and Startups' "2026 R&D Performance Dissemination Smart Factory Build Support Project." Under the project, Ivex will build an AI-based inspection system and a logistics automation system for Husteel Precision's hot-forging scroll production site.
The project is intended to apply national R&D achievements in smart manufacturing from the past 5 years to small and mid-sized manufacturing sites. Rather than developing new technologies, it focuses on implementing automation equipment, ICT, and AI solutions based on previously secured R&D outcomes and spreading them on-site.
Ivex carried out the Ministry of Trade, Industry and Energy's materials and components technology development project from 2021 to 2023, during which it developed a deep learning inspection solution for smart factories. For this project, it will apply manufacturing data collection and processing, deep learning-based automated inspection, and optical and inspection system technologies, along with MLOps, unsupervised anomaly detection, and robot vision technologies.
The target application is a hot-forging scroll for automotive compressors. Because the scroll has complex and deep structures, including spirals, it is difficult to reliably inspect internal areas with fixed machine vision alone. In addition, lighting reflections and shadow conditions vary depending on changes in the product's position and orientation.
A logistics automation task is also part of the effort. The product goes through heat treatment, shot blasting, and machining processes, during which its alignment repeatedly becomes disrupted. As a result, the downstream process requires technology that can quickly identify the position and orientation of misaligned products and bin-pick them.
To address this, Ivex implemented a structure that links vision AI, AI robotics, and MLOps. The platform consists of AIVision, AIVot, and AIVOps. AIVision recognizes the position and orientation of misaligned scrolls, and AIVot bin-picks the products.
AIVot then adjusts the product's position and orientation so that the internal inspection surface of the bin-picked product is clearly visible to the optical system, and acquires images under optimized conditions. AIVision then analyzes the captured images and determines quality.
Images generated during the inspection process and the AI judgment results are accumulated in AIVOps. The accumulated data are used to retrain models in response to new defects and changes in production conditions, and are also used for version management and performance management.
Ivex sees the structure as a "Closed-loop" physical AI system. The system flow runs from recognition to bin-picking, inspection posture control, quality judgment, and data accumulation. The key is to connect robot control, inspection, and AI operations in a single flow beyond the level of defect-judgment inspection equipment.
The company does not plan to limit the technology to hot-forging scrolls and intends to expand its application scope. The target applications are expected to include complex 3D-shaped parts and misaligned parts. Examples of sectors under consideration include automotive forging, casting, and machining, as well as secondary batteries, electronics, food, shipbuilding, semiconductors, and defense. The company believes these sectors have demand for inspection, logistics, and assembly automation.
Seong Min-su, CEO of Ivex, said that a single high-accuracy AI model is not enough to complete the AI transition in manufacturing sites, and that an organic connection among optics, robots, vision AI, and MLOps is necessary. He added that, based on its accumulated technologies integrating vision AI, robot control, optics, and MLOps, Ivex will push ahead with an AI-centered transformation of manufacturing processes that previously required direct human checking, picking, and judgment.
Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=409310
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.