Human skills data
Teaching machines how humans interact with the real world.
Advanced robotics and embodied AI need data that represents human actions, tasks, decisions and interactions with physical environments.
What we collect
Signals, structured.
Task demonstrations
Step-by-step execution of physical and digital tasks under a defined brief.
Object interaction
How humans grasp, manipulate, place and use tools and objects.
Movement sequences
Body and hand movement patterns captured across repeated attempts.
Environment interaction
Navigation and interaction within homes, workplaces and specialised settings.
Potential applications
- Robot manipulation and policy learning
- Embodied agents and world models
- Procedural understanding and planning
- Human-robot interaction research
Collection workflow
- 01Task specification agreed with your team, including success criteria.
- 02Contributors briefed with instructions, viewpoints and capture setup.
- 03Sessions recorded through controlled collection workflows.
- 04Segments labelled and reviewed against the specification.
- 05Approved data structured and prepared for delivery.
Building an AI system that needs better data?
Tell us what your model needs to learn. We'll explore how a purpose-built human data program could support it.
Physical AI
Teaching machines how humans interact with the real world.
Advanced robotics and embodied AI require data that represents human actions, tasks, decisions and interactions with physical environments — not just text scraped from the web.
- 01
Human Action
- 02
Capture
- 03
Structured Data
- 04
Model Training
- 05
Physical AI