Axonnic Launches EMG Sleeve for Robot Training

Axonnic is betting that robot learning needs richer demonstrations, not just more cameras. The approach is promising, but production value will depend on whether EMG meaningfully improves policies beyond vision, force sensing, and instrumented gloves.

• –32-channel EMG captures effort and movement signals cameras cannot directly observe.
• –Aligned EMG, video, and pose data could improve training for contact-rich manipulation.
• –The sleeve is used during data collection; deployed robots operate with their own sensors.
• –Early demos show real-time finger decoding and robotic-hand control, but published task-level results remain limited.
• –Its strongest wedge is force-sensitive industrial work where small grip errors can ruin the outcome.