Telemetry Correlation Debugging

2. Normalize timestamps before matching

Do not compare local wall time to robot logs directly.

Use one shared time domain:

• Prefer UTC if both sources expose wall-clock timestamps.
• Prefer epoch seconds if the VCP capture includes them.

For BiBa telemetry sessions, the common pattern has been:

• local workstation log: +01:00
• robot shell clock: +03:00
• container logs: UTC Z

If these domains are not stated explicitly in the result, the comparison is incomplete.

3. Match event plateaus, not individual samples

The robot battery log is typically much coarser than the VCP stream. Treat the robot log as sparse checkpoints.

Match using event shape:

• current magnitude plateau, such as 3.00 A
• battery direction transition, such as IDLE -> DIS -> IDLE
• voltage step accompanying the current event
• ordering of plateaus, such as 3.0 -> 4.9 -> 3.2 A

Do not claim exact point-by-point equality when one side only logs every few seconds.

4. Separate valid conclusions from invalid ones

Valid conclusions:

• both sides saw the same sustained DIS event
• app-side event appears about N seconds after the robot-side log point
• robot-side logging frequency is too coarse to capture short peaks

Invalid conclusions:

• the robot never produced a short current plateau just because it is absent from a coarse log
• app-side latency is exactly N seconds from one sparse match
• the transmitter display is wrong merely because it shows more detail than the robot log

5. Report both the match and the gap

Every comparison should include:

• matched plateaus
• unmatched short events
• observed lag range
• confidence level based on sampling density

If the robot logs once every about 5 seconds, explicitly say that short events between samples can be missed.

Quick Commands

Typical local VCP inspection:

grep -n 'RAWI=\|RAWDIR=' artifacts/telemetry-captures/vcp-YYYYMMDD-HHMMSS.log

Typical robot log extraction:

sshpass -p 'open' ssh -o StrictHostKeyChecking=no [email protected] \
	"docker logs --timestamps biba-biba-controller-1 2>&1 | grep 'Battery telemetry' | tail -n 120"

Typical narrow robot time window:

sshpass -p 'open' ssh -o StrictHostKeyChecking=no [email protected] \
	"docker logs --timestamps biba-biba-controller-1 2>&1 | grep '2026-03-28T20:53' | grep 'Battery telemetry'"

Output Pattern

Use a compact table when presenting results:

Then summarize:

1. what clearly matches
2. what the coarse robot log missed
3. what latency range is supportable
4. what additional logging frequency would be needed for tighter proof

Common Mistakes

BiBa-Specific Notes

• The active controller container may be biba-biba-controller-1, not biba-controller.
• VCP capture files can begin with one truncated line if capture starts mid-stream.
• App-side logs can show finer detail than robot logs; that is expected when the transmitter logs more frequently.