Leader ↔ follower teleop

The HMI has two teleop modes — leader↔follower (this page) and human-pose teleop (drive both arms from your webcam). They're mutually exclusive — only one runs at a time across the whole HMI.

Leader ↔ follower (60 Hz)

Turns the HMI into a leader/follower bridge between the two physical arms.

1. On the dashboard, find the Teleop card.
2. Pick the arm you'll back-drive in the Leader dropdown.
3. Pick the arm that should mirror it in the Follower dropdown (or click ⇄ to swap).
4. Click start. The backend:
   • disables torque on the leader (so you can move it freely),
   • enables torque on the follower,
   • spawns a thread reading the leader's positions, clamping to the follower's calibrated limits, and writing them as the follower's goal at 60 Hz.
5. Move the leader arm by hand. The follower mirrors. The header chip on each arm card flips to TELEOP · LEADER / TELEOP · FOLLOWER (green), and joint sliders / Home / preset replay are disabled for the duration.
6. Click stop to end. Both arms restore to manual with torque on.

E-STOP also stops teleop before disabling torque — so the follower can't jump to a stale queued goal when you next re-engage.

Midpoint mismatch

If teleop tracking shows a per-joint offset (most visibly on shoulder_lift, which changes arm height), the two arms were calibrated in different physical neutral poses. Fix with the calibration wizard: re-calibrate one arm while it holds the other arm's neutral pose. Or see the SO-101 arms guide for the underlying issue.

REST surface

See HMI overview for the full endpoint table.

See also

• Human-pose teleop — drive both arms bimanually from a laptop webcam (mutually exclusive with this mode).
• Calibration wizard — fix midpoint mismatches without leaving the browser.
• Design spec — full HMI design including teleop frame schemas.