Tutorials

All tutorials are jupyter notebooks, downloadable from https://github.com/MarcToussaint/rai-tutorials. For installation instructions, see Getting Started.

The first 3 are intro tutorials that introduce essentials for understanding the code:

• Configurations: The core data structure used to represent scenes and robots, compute features, and feed into optimization problems, simulations, and real robot control.

• BotOp: The Robot Operation interface used to control a real or simulated robot, as well as to access sensors.

• KOMO: A framework to formulate optimization problems, esp. for motion design (IK, path optimization, and manipulation planning).

• Intro: Configurations
• Intro: BotOp (Robot Operation) interface
• Grasp Test
• Intro: KOMO - Motion Optimization

The remaining tutorials cover various topics in more depth:

• BotOp-2: Real robot operation checklist & first steps
• Config-2: Computing differentiable features & collision evaluation
• Config-3: Importing, editing & manipulating them
• KOMO-2: Reporting & explaining convergence
• KOMO-3: Manipulation Modelling & Execution
• LGP-1: First Mini Interface
• Extension - Simulation: Low-level stepping interface & gym environments
• Extension - Rendering: Basic opengl, offscreen (headless), and interface to physics-based rendering
• Extension - RRT: basic finding example
• Extension - NLP interface: Low-level NLP formulation and solving
• Extension - Gym Environment Interface: minimal example