Skip to content

MOVIN TRACIN for Robot Motion Data

Training data from where the work happens

Capture real human motion in homes, offices and outdoor sites, delivered in the format your robots train on.

Working with the teams building Physical AI

  • NVIDIA
  • Diden Robotics
  • Boston Dynamics
  • Robotics and AI Institute

Why MOVIN TRACIN

Robot-ready motion, without the mocap stage

Physical AI runs on human motion data, but most of it was built for animation: wrong skeletons, studio-only, hard to scale. MOVIN TRACIN fuses LiDAR and camera data to capture precise 3D human motion in the real world, and delivers it in the format your robots train on.

  • Accuracy

    LiDAR + camera fusion delivers high-fidelity 3D motion, markerless and precise.

  • Adaptability

    Training-ready output that plugs directly into Isaac Sim, Isaac Lab, MuJoCo and ROS.

  • Scalability

    Capture in-the-wild motion data beyond the studio, quickly and efficiently.

Works withNVIDIA SOMAIsaac SimIsaac LabMuJoCoROSand more to come
MOVINNVIDIA

Launch partner for NVIDIA SOMA

The only team in Korea with early access to NVIDIA SOMA, the standard human model for Physical AI. Real-world human motion, ready for Isaac Sim, Isaac Lab and MuJoCo.

Built for robot learning

Real human motion data, ready to train on

Synced exocentric data

Motion, LiDAR depth and RGB, captured together in one frame. Accurate, fully 3D and grounded in the real world.

Motion, point cloud and RGB, time-synced
3 streams
Deviation vs optical reference
2–3 cm

Retargeted to your robot

Data arrives mapped to your robot’s joints. No hours of cleanup, no mocap expertise needed.

NVIDIA SOMA standard skeleton
77 joints
Delivered fitted to your robot’s structure
0 cleanup

Scalable data collection, anywhere

No studio, suit or markers. Anyone can run it after a short walkthrough, so you can scale across sites and collect in parallel. More rigs, more data.

Calibration per performer
3 sec
From unboxing to capture
3 min

Real tasks. Real environments.

Everyday work captured on location, shown as 3D motion.

  • IndoorSweeping the floor of a large room

  • OutdoorBending down to clean an outdoor parking lot

  • OfficeHandling different cleaning tools

  • Tight spaceCleaning a toilet and ledge

Built for every Physical AI team

Human motion infrastructure, from the lab to the field.

  • Humanoid robot makers

    Datasets retargeted to your own robot’s morphology, for walking, balance and whole-body control.

  • Research labs

    From capture to Isaac Lab or MuJoCo in one session. No stage to book.

  • World model and embodied AI teams

    Multimodal motion, point cloud and RGB, aligned in time and space.

  • Data operations

    Lightweight rigs anyone can run, so collection scales across sites.

Your setup

Everything you need to start collecting.

Proven with robotics partners

MOVIN works directly with the teams building humanoid robots, from simulation platforms to full-stack robot makers.

“Working with MOVIN, we will secure data that fits our robot hardware and strengthen a learning framework we can apply directly to real robot development.”
Junha Kim, CEO, Diden Robotics

Build your training dataset

Tell us your tasks, environment and target robot. We’ll scope the capture.