Metadata-Version: 2.5
Name: bounded-embodiment-kit
Version: 0.1.0
Summary: The lab bench for the Bounded Embodiment tests: adapters, instruments, the runner and its watchdog.
Project-URL: Repository, https://github.com/RobotRegistryFoundation/bounded-embodiment
Author: Craig Merry
License-Expression: Apache-2.0
Requires-Python: >=3.10
Requires-Dist: bounded-embodiment~=0.1.0
Requires-Dist: numpy>=1.24
Provides-Extra: camera
Requires-Dist: opencv-python-headless<4.14,>=4.9; extra == 'camera'
Provides-Extra: fit
Requires-Dist: opencv-python-headless<4.14,>=4.9; extra == 'fit'
Requires-Dist: scipy>=1.10; extra == 'fit'
Provides-Extra: lerobot
Requires-Dist: lerobot>=0.6.1; (python_version >= '3.12') and extra == 'lerobot'
Requires-Dist: pyserial>=3.5; extra == 'lerobot'
Provides-Extra: oakd
Requires-Dist: depthai>=2.24; extra == 'oakd'
Requires-Dist: opencv-python-headless<4.14,>=4.9; extra == 'oakd'
Provides-Extra: test
Requires-Dist: pytest>=8; extra == 'test'
Description-Content-Type: text/markdown

# bounded-embodiment-kit

The lab bench for the Bounded Embodiment tests: adapters (LeRobot SO-10x, a Python template,
simulators), the runner and its watchdog (`be run`, `be goto-start`, `be doctor`), the OAK-D recorder
(`be record`), the markerless tracker (`be fit`), the accuracy check (`be instrument`), and packaging
(`be report`, `be submit`). It adds these commands to core's `be`.

**The adapters, runner and watchdog are a draft that needs line-by-line human review before they
drive hardware.** The markerless tracker is validated on synthetic recordings only. Method v0.1 is a
draft for comment; a verdict is never a certification.

Extras: `fit` (the tracker: OpenCV, SciPy), `oakd` (DepthAI), `camera` (OpenCV), `lerobot`.

Start with the lab guide: https://github.com/RobotRegistryFoundation/bounded-embodiment/blob/main/docs/lab-guide.md
