[For more corporate information on the startups featured in this article, please visit the Unicorn Factory big data platform ‘Data Lab’.]
Ultra-low-power AI semiconductor startup DeepX announced on the 6th that it is embarking on the development of an “AI (artificial intelligence) welding robot” to be utilized in nuclear power plant manufacturing sites, in collaboration with Doosan Robotics.
This development project will proceed through cooperation among four institutional investors: Doosan Robotics, DeepX, Sage, and the Industry-Academia Cooperation Foundation of Changwon National University, under the Ministry of Trade, Industry and Energy’s “2026 Robot Industry Technology Development Project.” The consortium plans to develop technology that uses AI to recognize and correct welding targets and paths, analyzes vision and thermal imaging data during the welding process, and determines welding status and anomalies. Subsequently, they aim to implement this as an AI welding robot solution by integrating it with collaborative robots.
DeepX will develop NPU (Neural Processing Unit)-based equipment and a Software Development Kit (SDK), build a 3-channel vision data pipeline, integrate with the robot controller, develop thermal imaging analysis AI models, and verify environmental durability in welding environments.
Doosan Robotics, the lead institutional investor, is responsible for system integration design and will develop collaborative robot welding cells, micro-displacement correction AI, and real-time welder control technology. Sage will develop 3D vision-based recognition of welding targets, welding path correction, and an AI model for detecting welding bead anomalies, while the Changwon National University Industry-Academia Cooperation Foundation will develop AI-based welding path generation and physics-law-based digital twin technology.
During the project period, they plan to conduct phased field validation at Doosan Enerbility’s nuclear power plant equipment manufacturing sites. The goal is to complete quality verification of nuclear power plant equipment by 2029 and begin commercialization from 2030.
DeepX stated, “This project will serve as a concrete application case where ultra-low-power AI semiconductors are combined with robots and utilized in actual manufacturing sites,” adding, “We will build an AI computing environment capable of efficiently processing the AI computations required in manufacturing sites and operating stably.”
[MoneyToday Startup Media Platform Unicorn Factory]
in FUTURE
