motor.set-zero-pos

Set a motor’s zero-position reference to its current mechanical position.

Sets the motor’s zero-position reference to its current mechanical position (persisted to flash). Damiao motors require a power cycle afterward.

Each motor’s encoder zero is calibrated at one of the joint’s mechanical end stops, not at the robot’s rest position. In --guided mode the CLI walks every joint to an end stop and zeroes there; in single---id mode you are responsible for holding the joint at the intended end stop before zeroing. AxolArm carries a per-joint offset so the public API stays in joint frame (0 = rest).

Flag	Description
--l / --r	Select the left or right arm (mutually exclusive, required)
--channel IFACE	SocketCAN interface to use instead of the selected arm’s Axol hub interface, for setups without the Axol hub CAN adapter (e.g. can0)
--id ID	CAN ID of the motor to zero, hex or decimal (required unless --guided)
--type {myactuator,damiao}	Motor type (inferred from --id if omitted)
--guided	Walk through the arm joints, zeroing each at an end stop
--joints j1,j2,…	Subset of arm joints to walk in --guided mode (default: all seven; the gripper has no zero — it self-calibrates at enable time)

In --guided mode the CLI first asks you to hold every selected joint somewhere inside its operating range (away from the end stops) and press Enter once — it captures each joint’s in-range start reference concurrently. It then walks the selected arm joints one at a time (the gripper is auto-calibrated at runtime): for each, move it to the prompted end stop and press Enter. If the motion is in the expected direction the zero is committed immediately. If you moved the wrong way, the CLI keeps your starting reference and asks you to travel to the other end stop instead; pressing Enter there sets the zero. Press Ctrl-C to skip a joint.

wrist_2 and wrist_3 are special: their end stops are the only ones that are not laser-aligned, so per unit one stop can be better placed than the other. For these two joints the guided flow accepts either end stop — move to whichever you trust and press Enter. Which side a robot was zeroed at is detected automatically at runtime from the encoder reading (zeroed at the upper stop, all readings are ≤ 0; at the lower stop, ≥ 0), so no side has to be recorded anywhere and robots zeroed by the older forced-side flow keep working unchanged. The only requirement: don’t park these joints exactly at an end stop when connecting, since that one position is ambiguous.

axol motor.set-zero-pos --l --id 0x01                        # single motor
axol motor.set-zero-pos --l --guided                         # all left-arm joints
axol motor.set-zero-pos --l --guided --joints wrist_2,wrist_3  # just the wrist Damiaos

After --guided calibration each motor’s encoder zero coincides with its calibration end stop. AxolArm carries a per-joint offset internally so the public API (positions, motion_control, etc.) stays in joint frame (0 = rest position); for wrist_2/wrist_3 the offset’s sign is detected from the encoder at connect time. Damiao motors (WRIST_2, WRIST_3) need a power cycle for the new zero to take effect.

Bring-up refuses a robot whose zeros were never set. An unset (or stale) encoder zero makes every joint-frame reading garbage, so enabling the arms — from teleop, collect-data, run-policy, gravity-comp, the control panel, or the diagnostics dashboard — first verifies each fixed-stop joint’s reading is plausible for a zero at its calibration end stop, before any motor is torqued. If one isn’t, it stops with an error naming the joint; run axol motor.set-zero-pos --guided to (re)zero, then try again. (wrist_2 / wrist_3 get the equivalent check from their end-stop side detection.)