Brief the operator on the lineage's measured zero-command and untrained-axis behavior before handing over the joystick

Symptom

A teleop session was about to start on a policy that does not stand still at zero command and has never been trained on lateral commands - behaviors an unbriefed operator would report as bugs or emergencies.

Context

Three measured facts were written into the teleop instructions ("都有 实测依据, 不是猜"): (1) A/D (lateral) keys will get essentially no response - probe-measured sidewalk tracking ~3%, an untrained axis: "这正是 C4 要解决的事, 不是 bug"; (2) no keypress = cmd 0, and this lineage does not stand still at zero command - a three-generation lineage property: paces in place, drifts right ~5 cm/s, net rotation -30 deg/20 s; sim survival is 20/20 (it will not fall) but it walks away slowly, so leave floor margin especially on the right; (3) S (backward) WILL respond - probe-measured 20/20 survival, 67% tracking untrained, which is also why this root was chosen for the C ladder. Plus a keybinding dry-run while suspended before touching down.

Change

Operator briefing became part of the deployment artifact: expected response per key, expected idle behavior with magnitudes and directions, and the distinction between untrained (expected, not a bug) and abnormal.

Outcome

The session proceeded with correct interpretations available in advance; the known zero-command wander was handled by floor margin and start-with-command procedure rather than misdiagnosed on the spot.

Mechanism

A learned policy's off-nominal behaviors (idle drift, untrained axes) are lineage properties, stable and measurable in sim beforehand; operator surprise converts known properties into false incident reports and unsafe reactions. A briefing transfers the measured behavior model to the person holding the controller.

Applies when

• handing a learned policy to an operator or demo audience
• the policy idles in a non-stationary way at zero command
• some command axes are untrained in the current lineage