sEMG-based third arm
Reconstructing the full 3D pose of the hand from forearm EMG, then using it to control an extra robotic limb, not just a prosthetic.
2026–27In progressEMGMLHardware
The idea
This project builds on the EMG-to-hand-pose work from last year’s ASL, prosthetics and piano teams. The question: using MindRove EMG sensors placed away from the wrist, can we reconstruct a continuous 3D hand pose, rather than only classifying a small set of gestures?
The team will then use that control for a third limb: an extra robotic arm worn alongside your own two, rather than only a replacement prosthetic. For background, see this research paper on supernumerary robotic limbs.
What the team will do
- Collect synchronized EMG and hand-pose data
- Train models that reconstruct individual joint angles from EMG
- Test how well they hold up across recording sessions and electrode placements
Later this year or in 2027–28, this can grow into real-time control of a fully dynamic prosthetic hand.
How we’ll measure it
- Dropped or crushed objects
- Grip-force error
- Task completion time
- Performance without visual feedback
- Discrete versus proportional EMG control