Tutorials

Tutorials#

These tutorials walk you through EmbodiChain step by step, from creating your first simulation scene to training RL agents. Each tutorial includes a complete runnable script and a line-by-line explanation.

Suggested Learning Path#

Follow the tutorials in this order for the best learning experience:

Phase 1: Simulation Basics

  1. Creating a simulation scene — Set up a simulation, add objects, and run the render loop. Start here.

  2. Creating a soft-body simulation and Creating a cloth simulation — Add deformable bodies to your scenes.

  3. Rigid object group tutorial — Manage collections of rigid objects efficiently.

  4. Rigid constraint tutorial — Attach and detach two rigid objects via a fixed constraint.

  5. Simulating a Robot — Load and control a robot in simulation.

  6. Simulating a Camera Sensor — Add cameras and capture RGB/depth/segmentation data.

  7. Create a solver — Configure IK solvers for end-effector control.

  8. Motion Generator — Generate smooth trajectories with motion planners.

  9. Atomic Actions — Use built-in action primitives (move, move joints, pick, move held object, place).

  10. Interactive Robot Control with Gizmo — Interactively control robots with on-screen gizmos.

Phase 2: Environments

  1. Creating a Basic Environment — Create a simple Gymnasium environment with BaseEnv. Prerequisite: Phase 1 basics.

  2. Creating a Modular Environment — Build a config-driven environment with EmbodiedEnv, managers, and randomization. Prerequisite: Creating a Basic Environment.

  3. Data Generation — Generate expert demonstration datasets for imitation learning. Prerequisite: Creating a Modular Environment.

  4. Reinforcement Learning Training — Train RL agents with PPO or GRPO. Prerequisite: Creating a Basic Environment.

Phase 3: Extending the Framework

  1. add_robot — Add a new robot model to EmbodiChain.