docs(feetech): documenting root cause of wrist calibration issues#3348
docs(feetech): documenting root cause of wrist calibration issues#3348Maximellerbach wants to merge 3 commits intomainfrom
Conversation
github-actions
bot
added
the
documentation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
There was a problem hiding this comment.
Pull request overview
Adds Feetech documentation to help users diagnose and mitigate wrist calibration/position-reading issues, and updates the docs navigation entry accordingly.
Changes:
- Adds a new “Troubleshooting → Position Overflow” section with a quick recovery procedure and an offset-reset walkthrough.
- Renames/restructures headings in the Feetech firmware update guide for clearer hierarchy.
- Updates the docs sidebar title for the Feetech page.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| docs/source/feetech.mdx | Introduces troubleshooting guidance for position overflow and refactors section headings. |
| docs/source/_toctree.yml | Updates the displayed sidebar title for the Feetech documentation page. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| The firmware may be overflowing and returning incorrect position readings. | ||
|
|
||
| **Quick fix:** Unplug both the AC power and USB cable, then reconnect them. This should give you correct position readings again. |
There was a problem hiding this comment.
The PR description mentions a specific diagnosis (homing offset + current position overflow beyond the [0, 4095] range after consecutive calibrations), but the doc here only says the firmware “may be overflowing”. Consider documenting the actual suspected mechanism and the expected valid position range so readers can verify they’re hitting the same issue (and distinguish it from other calibration failures).
docs/source/feetech.mdx
Outdated
| 2. Select the *Programming* tab | ||
| 3. Move all joints to the middle of their range | ||
| 4. Click *Offset* | ||
| <img | ||
| src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/lerobot/feetech_reset_offset.png" | ||
| alt="Feetech Offset Position" | ||
| /> |
There was a problem hiding this comment.
The ordered list ends immediately before the block tag without a blank line/indentation, which can break list rendering in MDX/Markdown parsers. Align this with the pattern used elsewhere in the docs (e.g., docs/source/so100.mdx inserts a blank line before images after list items) by adding a blank line and/or indenting the image block under the relevant list item.
| - local: feetech | ||
| title: Updating Feetech Firmware | ||
| title: Feetech Troubleshooting and Firmware Update |
There was a problem hiding this comment.
The sidebar title (“Feetech Troubleshooting and Firmware Update”) doesn’t match the page H1 (“Feetech Troubleshooting and Motor Firmware Update”). Please make these consistent to avoid confusion/search mismatches in the rendered docs navigation.
Adding documentation about a bug with the feetech motors
Type / Scope
Summary / Motivation
We encountered negative positive values coming back from the feetech motors, which is not normal.
The diagnosis is that when doing multiple consecutive calibration, it may be possible that the combination of homing offset + current position overflows the [0,4095] range.
While we find a proper fix, we decided to document the problem.
We found that unplugging the robot (AC and USB) and plugging it back solved the overflow problem in our case.
Related issues
calibration failed : ValueError: Magnitude 2114 exceeds 2047 (max for sign_bit_index=11) #1296
Fix wrist roll calibration #3196
fix: symmetric calibration for wrist_roll motors in SO-101 #3320
so101 wrist_roll calibration broken in new version: leader/follower zero points don't correspond #3193
Checklist (required before merge)
pre-commit run -a)pytest)Reviewer notes
We will probably need to find a more long term fix