Checklist v3: IMU calibration (level twice, straight pull-aways and stops); yukart/raw/imu

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DDwfd6Ki8UfG4hK6W5YpJz

docs/impl/plans/Z27AG-CAPTURES.md

@@ -1,7 +1,7 @@
 ---
 type: plan
 status: ready
-updated: 2026-10-13
+updated: 2026-10-14
 ---
 # Z27AG decoding captures: what to do in the car
 
@@ -17,6 +17,7 @@ ground truth. Prior art and first candidates: [the prior-art report](../reports/
 | `vehicle.ndjson` | `decoder` and `obd-poller` via the bus | decoded CAN signals, and the `obd_*` references from the K-line through the ELM327 (rpm, speed, throttle, load, MAF, timing, O2 every round; coolant, intake air, trims every 5 s); `env.timestamp` is the reply's arrival at the Orin |
 | `events.ndjson` | marks from the app | when each step happened |
 | `gnss.ndjson` | gpsd (SIM7600) | speed and position for the drive |
+| `imu.ndjson` | `imu` via the bus (`yukart/raw/imu`) | every IMU sample, raw module axes, Orin receive time: calibration (27–30) and, once calibrated, motion for the drive |
 
 ## Operating it from the app
 
@@ -53,14 +54,20 @@ recorder refuses marks with "no session" otherwise); engine off, doors shut.
 - **Engine running, parked** (14–18): start and idle, held revs in neutral (no OBD
   on this car's CAN: the revs are approximate, read off the tacho), blips, A/C load, radiator fan.
 - **Driving** (19–22): steady speeds, stops, corners, each gear. A passenger presses,
-  or skip the marks: GNSS speed is recorded continuously, and gear comes from
-  rpm ÷ speed.
+  or skip the marks: GNSS speed is recorded continuously. The gear is not on CAN and
+  is not inferred (decode report).
 - **On jack stands** (23–26, since v2): front wheels up, rears chocked, handbrake
   off. Unpowered steering; speedometer readings held in 2nd and 3rd (fronts turn,
   rears stay still, which tells the wheels apart); braking from speed with vacuum
   assist; reverse. Stands replace the driving speeds where a drive is not practical.
+- **IMU calibration** (27–30, since v3): once the IMU is fixed in its final place. Level
+  twice on the same spot, facing one way and then turned round (a rough 180° is fine:
+  the ground's slope reverses and cancels, the mounting's tilt does not); then three
+  firm pull-aways and three firm stops on a straight, flat section (forward, signed by
+  which kind of stretch it is). Any session holds raw IMU samples (`imu.ndjson`), so
+  these can go in their own short session or with anything else.
 
-Version 2 (2026-10-13) adds the stands section and writes down the conditions the
+Version 3 (2026-10-14) adds the IMU calibration section. Version 2 (2026-10-13) adds the stands section and writes down the conditions the
 first run showed matter: engine off and handbrake on for 1–13, foot brake held for
 the handbrake step, neutral with the clutch up for the revs.
 
@@ -97,6 +104,17 @@ on the Orin), and the signal list's `elm_connected` shows the Bluetooth link:
 
 ## Afterwards
 
+**IMU calibration**, on the car Orin, from the session that holds steps 27–30:
+
+```bash
+yukart-imu --calibrate /var/lib/yukart/recordings/<session>           # print: down, forward, slope, spread
+yukart-imu --calibrate /var/lib/yukart/recordings/<session> --write   # and save /var/lib/yukart/imu/mount.json
+```
+
+The service picks the new mounting up within seconds; roll and pitch should then read
+about 0 parked on level ground. Steps 27–28 alone also work: forward then stays the
+`--forward` axis, laid flat, until a session with 29–30 is calibrated.
+
 Sessions are in `/var/lib/yukart/recordings/` on the car Orin. Copy one off with
 `rsync -a mikkeli-orin-nano:/var/lib/yukart/recordings/<session>/ …` into
 `yukart-data`, with a line in its README of what was done.