Welcome to bard’s documentation!

bard is an efficient, PyTorch-native library for rigid-body dynamics, designed for batch processing and differentiability. It’s an ideal tool for robotics research, especially in areas like reinforcement learning, trajectory optimization, and model-based control.

Contents:

• Installation
  • CUDA Installation (Recommended for GPU Acceleration)
  • CPU-Only Installation
  • Upgrading from CPU to GPU
  • Developer and Test Setup
  • Building the Documentation
• Quick Start
  • A Simple Example
  • Loading a URDF from a File
  • Enabling Compilation for Maximum Performance
• API Reference
  • Robot Chain
  • Kinematics
  • Jacobian
  • Dynamics
  • URDF Parsing
  • Data Structures