API Reference

This section provides the API-level documentation for EmbodiChain’s public Python modules.

Use this reference when you need:

• module-level overviews and responsibilities,

• public classes, functions, and configuration objects,

• links into specialized subpackages (simulation, gym environments, RL, and utilities).

The pages are organized from high-level package namespaces to concrete submodules.

Core Framework

The core embodichain framework is split into six top-level packages:

compute

Shared numerical computations: analytical kinematics, trajectory interpolation/resampling/warping, geometry queries, and image layouts.

data

Dataset resolution, asset-download helpers, shared constants, and enums used by simulation tasks and training pipelines.

data_pipeline

Online data streaming and recording: a process-safe trajectory engine, online datasets/samplers for live-simulation training, and compressed depth-sidecar storage for LeRobot datasets.

lab

The simulation laboratory: the sim core (scene, objects, sensors, IK solvers, planners, atomic actions), the gym environment framework, real-device controllers (devices), and browser visualization.

toolkits

Standalone asset-preparation and manipulation utilities (parallel-gripper grasp sampling, URDF convex decomposition, URDF assembly) usable independently of the simulation loop.

learning

Learning systems, currently the rl subpackage: on-policy RL algorithms (PPO/GRPO), rollout buffers, collectors, policy/model builders, and the training entry point.

utils

Shared utilities: the @configclass decorator, logging, math/tensor helpers, file/string/device helpers, and legacy computation import aliases.

compute	Array and tensor computations shared across simulation, planning, and tools.
data	Dataset resolution, asset-download helpers, preset asset registries, and shared constants/enums used by simulation tasks and training pipelines.
data_pipeline	Online data streaming and recording for live-simulation training.
lab	EmbodiChain's robotics laboratory.
toolkits	Standalone asset-preparation and manipulation utilities usable independently of the simulation loop.
learning	Learning systems.
utils	Shared utilities used across EmbodiChain.

Public API Coverage

Public Python APIs are declared through static __all__ values in non-private modules. Curated API pages remain the preferred place for explanations and examples. The fallback supplement keeps less prominent exports visible through their signatures and source docstring summaries.

Run the read-only checker after changing module exports or API docs:

python docs/scripts/check_api_docs.py

The checker never edits repository files. If it reports missing exports, use the /update-api-docs agent skill to add the appropriate API entries and documentation. CI runs this same checker after style checks and before tests.

• Public API Supplement