Config, a Seoul and San Jose-based startup, launched a robotics data platform designed to supply curated motion datasets for robot foundation models, featuring a proprietary data conversion process that adapts human-recorded motion for robotic systems. Samsung Venture Investment led the company’s $27 million seed round alongside venture arms from Hyundai, LG and SKT, backing a startup founded in 2025 by former Meta researcher Minjoon Seo and engineers with backgrounds at companies including Waymo, Google and Naver.
The company records human task performance in studio and real-world environments across Seoul and Hanoi, supported by a workforce of nearly 300 focused on large-scale motion data production. Config said it has accumulated more than 100,000 hours of human motion data and differentiates itself by transforming datasets before model training so the resulting AI systems better align with robotic movement and physical interaction requirements.
Config’s approach also reflects a broader industry trend toward specialised infrastructure providers supplying foundational datasets and tooling for physical AI and robotics development.
Robotics Data Conversion Services
Config Launches Its Robotics Data Platform
Trend Themes
-
Human-to-robot Motion Conversion — A proprietary conversion layer that maps human-recorded motion to robotic kinematics enables more realistic and safe robot behaviors in unstructured environments.
-
Curated Motion Dataset Market — Rising demand for large, labeled motion corpora tailored to physical constraints is creating value in differentiated datasets that improve model generalization for physical tasks.
-
Specialized Foundation Data Providers — The emergence of vendors focused on preprocessed, domain-specific data and tooling is reducing integration friction for robotics developers and shifting competitive moats to data quality and conversion IP.
Industry Implications
-
Manufacturing Robotics — Robots trained on converted human motion data could enable more adaptable assembly and collaboration with human workers on complex tasks.
-
Healthcare Assistive Robotics — Assistive devices and rehab robots informed by realistic human motion datasets may offer finer-grained, patient-specific interaction and movement support.
-
Logistics and Warehouse Automation — Automation systems leveraging curated motion primitives can improve handling of varied objects and dynamic human-robot workflows in fulfillment centers.