“피지컬 AI 검증, 기존 임베디드 검증 체계 그대로 적용해야”
기사입력2026.03.20 08:47‘e4ds Physical AI Frontier 2026’ 컨퍼런스에서 김민기 MDS테크 과장이 ‘피지컬 AI 개발을 위한 개발 전략과 실무 팁’을 주제로, 실제 현장에서 발생하기 쉬운 문제 지점과 이를 줄이기 위한 현실적인 접근법을 제시하며, 피지컬 AI 시대에는 AI가 코드를 자동으로 생성하더라도 물리적 사고와 비용 리스크를 줄이기 위해 개발자가 기능 안전 기준과 검증 체계를 중심으로 최종 판단과 책임을 맡아야 한다고 밝혔다.
댓글
관련 뉴스
[2025 e4ds Tech Day] Taejun Park, MDS Tech's Deputy Manager, "Providing an Easy-to-Use Environment for 'On-Device AI Development' Even Beginners Can Understand"
AI is now evolving into a technology that performs real-time inference on embedded devices rather than in the cloud. In this context, on-device AI faces high barriers to entry due to complex development environments, with particular difficulties in model optimization and debugging in MCU environments. Against this backdrop, MDS Tech is providing an end-to-end development workflow that enables even non-experts to apply on-device AI in practice, covering everything from model design, training, and optimization to MCU integration and debugging. In this regard, we met with Park Tae-jun, Deputy Manager at MDS Tech, who will present at the '2025 e4ds Tech Day' on September 9th, to discuss on-device AI development.
2025.08.12by 배종인 기자
.jpg)
[2025 e4ds Tech Day] "MCU-Based On-Device AI, Realized with 'TinyML' and Generative Platforms"
On-device AI development in MCU environments through 'TinyML,' an ultra-compact machine learning technology that enables AI models to run with memory ranging from tens of KB to several MB and power consumption in the milliwatt range, is gaining attention. At the '2025 e4ds Tech Day' event held on September 9th, Park Tae-jun, assistant manager at MDS Tech, presented on 'development solutions that break down barriers to entry for on-device AI,' providing detailed explanations on the definition of on-device AI, its advantages, market outlook, and changes in the development environment.
2025.10.20by 명세환 기자
MDS Tech Manager Kim Min-ki, “MDS Tech Supports the Full Lifecycle of Physical AI Embedded Development”
MDS Tech is a technology company that supports embedded system development across various industry sectors including automotive, defense, aviation, and semiconductors. The company is advancing developer environments through development and verification solutions including TRACE32. Based on its partnership with Arm, MDS Tech provides solutions covering the entire V-model lifecycle—including compilers, debuggers, and dynamic test tools—and goes beyond simple supply to offer education, technical support, and verification. Particularly in physical AI development, the company supports the entire platform development and verification process using Jetson, Keil MDK, TRACE32, and DT+Trace. In this context, MDS Tech plans to present on "Development Strategy and Practical Tips for Physical AI Development" at the 'e4ds Physical AI Frontier 2026' event scheduled for March 6. Accordingly, we met with Kim Min-gi, assistant manager at MDS Tech who will be giving the presentation, to discuss key points in physical AI design.
2026.02.23by 명세환 기자
“AI Now Moves”… 'e4ds Physical AI Frontier 2026' March 6, COEX
The 'e4ds Physical AI Frontier 2026' conference will be held on March 6th at COEX Rooms 317 and 318 in Samseong-dong, Seoul. This event is the nation's largest specialized conference dedicated to Physical AI, focusing on practical AI implementations that actually operate in industrial environments such as sensors, semiconductors, embedded systems, robotics, and manufacturing automation.
2026.02.24by 명세환 기자
.jpg)
“The Key to Applying 'Physical AI' to Industrial Fields Is the Combination of Experience Assets, AI, and Robotics Technology”
At the 'e4ds Physical AI Frontier 2026' event held on March 6th at the Seoul COEX Conference Room, Lee Soon-yeol, Chairman of the Korea Industry 4.0 Association, delivered a presentation on the topic of 'Implications of Physical AI Application in Industrial Settings' and presented key conditions for implementing Physical AI along with possibilities within Korea's industrial environment. Chairman Lee emphasized, "Physical AI that autonomously judges and acts in industrial settings is only completed when human experiential assets, artificial intelligence, and robotics technology are combined."
2026.03.16by 배종인 기자
