Mind and Motion Labs

The quality control and filtering layer for robotics training data

We automatically score and filter egocentric, VLA, and world-model training data so you can train on data you know is good, not data you hope is good.

Connect on LinkedIn Message on X

Tools & resources

Kriya Visualizer

Explore Action100M-style video annotations: timeline, segments, transcript. Load your own JSON + video or browse matched folders.

Try it GitHub
API Documentation

Integrate with our scoring and filtering APIs for robotics training data.

View docs
Kriya-Egocentric-100K

Action100M-style video action annotations for a preview subset of Egocentric-100K (Build AI). Hierarchical segments with LLM-generated descriptions.

Hugging Face
Kriya-EPIC-KITCHENS

Action100M-style annotations for a preview subset of EPIC-KITCHENS-100. Hierarchical temporal segments with LLM-written action descriptions.

Hugging Face
✓ Less manual review so your team can focus on robotics, not data janitorial work
✓ More usable data per collection run with automated filtering and scoring
✓ High-quality annotations and filtered trajectories for VLA and world models

Collaborating with leading robotics researchers in Singapore