Diden Robotics and Movin to Transfer 'Human Motion' Data to Humanoids

Diden Robotics, an industrial physical AI company, and Movin, a human motion AI company, have signed a Memorandum of Understanding (MOU) to build and utilize datasets required for humanoid learning.

The Diden Robotics and Movin logos next to a bipedal humanoid robot standing against a dark background, with the text "Diden Robotics and Movin Sign Technical Cooperation MOU to Build Humanoid Robot Learning Datasets"
Press

Diden Robotics, an industrial physical AI company, and Movin, a human motion AI company, announced on July 15 that they have signed a memorandum of understanding (MOU) to build and use datasets for humanoid learning.

Under the agreement, the two companies plan to improve robot motion learning performance together. They will combine Movin's LiDAR-based motion capture technology and multimodal data collection capabilities with Diden Robotics' humanoid development capabilities.

Improving humanoid performance requires precise data on how real people move, keep their balance, and interact with surrounding objects in complex environments.

Existing motion data is often built for content production or animation, which makes it hard to apply directly to a robot's joint structure, range of motion, and physical constraints.

The two companies will collect the diverse movements of people in real environments as high-precision 3D motion data and multimodal sensor data. They will then build a data pipeline together that converts this data into a form suited to humanoid robot learning.

Movin will collect LiDAR and camera data from multiple viewpoints and provide it retargeted to Diden Robotics' robot structure. Diden Robotics plans to use this dataset to raise the learning efficiency and precision of core behavior models for the humanoid, including walking, balance keeping, whole-body control, and interaction with objects and the environment.

Starting with this MOU, the two companies plan to expand their cooperation to joint research and the possibility of commercialization.

"Improving humanoid performance requires high-quality multimodal data that captures both human movement and the surrounding spatial information," said Joonha Kim, CEO of Diden Robotics. "Through our cooperation with Movin, we will secure data optimized for robot hardware and strengthen a learning framework that can be applied directly to real robot development."

"The technology to precisely collect human movement and convert it to fit a robot's body structure is an important foundation for humanoid robot learning," said Byeol Choi, CEO of Movin. "Through this cooperation, we hope Movin's motion data technology will expand beyond the content industry into core data infrastructure for robotics."

See how our robots
bring these capabilities to life