Map human motion onto a robot hand
Retargeting connects two different mechanisms. Begin by checking the mapping.
Explain why human and robot hand configurations need an explicit mapping.
Choose the embodiment first
Select a robot hand supported by the implementation you are using. A human hand and a robot hand may have different joints, ranges and coupling. The goal is a useful task motion, not necessarily identical joint angles.
Start with a documented example
Use the Dex Retargeting examples and their expected input format. Keep the first trial in visualization or simulation. AnyTeleop provides research context for connecting human input, a hand and an arm.
Check names, frames and limits
Match joints by name across the retargeter, robot model and driver. Check coordinate frames, left-versus-right configuration and motion limits. Verify one finger at a time before interpreting a complete grasp.
Share the failure as well as the fit
Record the model and code versions, input method and retargeting settings. Describe where contact or a pinch becomes inaccurate. Connecting real hardware requires a separate safe control and validation procedure; this guide is a research starting point.
Try it yourself
Your practical task
- Read the documented examples and identify a supported target hand.
- List its joint names, left/right configuration and input representation.
- Start with a visualized pinch; record fingertip mismatch before considering physical control.
What to produce: A mapping checklist and a clear description of one mismatch.
Check your understanding
Original sources & next steps
Dex Retargeting: maintained examples and requirements ↗AnyTeleop research ↗Original Dexhands teaching material. Interactive models and added references reviewed September 23, 2026; the introductory reading retains its September 22 review. Research links are not endorsements or evidence of hardware compatibility.