nvidia-isaac/nvblox

A GPU-accelerated TSDF and ESDF library for robots equipped with RGB-D cameras.

What it solves

nvblox provides a high-performance solution for real-time 3D scene reconstruction in robotics. It allows robots to map their surroundings in 3D, enabling them to understand geometry and navigate complex environments effectively.

How it works

The library performs 3D reconstruction directly on the GPU using NVIDIA CUDA hardware. It builds, manipulates, and queries spatial data structures in real-time, allowing for tasks like texture mapping and the generation of Euclidean Signed Distance Fields (ESDFs) for path planning.

Who it’s for

It is designed for robotics developers and researchers who need efficient, real-time spatial perception for autonomous systems. It supports integration through C++, Python, and ROS2.

Highlights

  • Real-time 3D reconstruction optimized for NVIDIA GPUs
  • Supports multiple interfaces: C++, Python, and ROS2
  • Capable of generating ESDFs for navigation and path planning
  • Handles complex tasks like multi-camera reconstruction and texture mapping

Related

  • Project
  • Project
  • Project
  • Project