feat: add script to segment lerobot dataset

[shuheng-liu]

What this does

Add functionality to segment one or more lerobot episodes into multiple episodes in a new dataset.

How it was tested

1. Ran CPU tests and

python src/opentau/scripts/segment_lerobot_dataset.py ~/.cache/huggingface/lerobot/physical-intelligence/libero /tmp/libero_segment /tmp/segments.json
python src/opentau/scripts/segment_lerobot_dataset.py ~/.cache/huggingface/lerobot/lerobot/droid_100/ /tmp/droid_segment /tmp/segments.json

where /tmp/segments.json reads

{"0": [[5, 15], [10, 20]], "1": [[0, 30]]}

How to checkout & try? (for the reviewer)

Run CPU tests and

python src/opentau/scripts/segment_lerobot_dataset.py ~/.cache/huggingface/lerobot/physical-intelligence/libero /tmp/libero_segment --episode-id 0 --segment 0:10 --segment 30:40 --segment 5:15
python src/opentau/scripts/segment_lerobot_dataset.py ~/.cache/huggingface/lerobot/lerobot/droid_100/ /tmp/droid_segment --episode-id 0 --segment 0:10 --segment 30:40 --segment 5:15

where /tmp/segments.json reads

{"0": [[5, 15], [10, 20]], "1": [[0, 30]]}

Checklist

• I have added Google-style docstrings to important functions and ensured function parameters are typed.
• My PR includes policy-related changes.
  • If the above is checked: I have run the GPU pytests (pytest -m "gpu") and regression tests.

[Copilot]

Pull request overview

This pull request adds a new script to segment a LeRobot dataset episode into multiple smaller episodes in a new output dataset. The implementation supports both v2.0 and v2.1 input formats and always outputs v2.1 format.

Changes:

• Added src/opentau/scripts/segment_lerobot_dataset.py script to segment episodes by frame ranges
• Added comprehensive test suite in tests/datasets/test_segment_lerobot_dataset.py with three test cases covering v2.1 input, v2.0->v2.1 conversion, and edge cases

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 5 comments.

File	Description
src/opentau/scripts/segment_lerobot_dataset.py	Main segmentation script with CLI argument parsing, dataset validation, parquet data slicing, metadata creation, and video/task handling
tests/datasets/test_segment_lerobot_dataset.py	Test suite with three comprehensive tests validating segmentation logic, version conversion, and edge cases like overlapping/non-consecutive segments

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[WilliamYue37]

Code looks good to me, but I think it could be better to have the user input a file with the segment start:end. For long recordings, adding an extra argument on the command line can get tedious/overwhelming.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[WilliamYue37]

Maybe we should have a segments.json example file in the examples folder.