nvidia-isaac/nvblox

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

1307 stars · 163 forks observed · 2026-09-23

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

nvblox is a GPU-accelerated C++/Python library for real-time 3D reconstruction using TSDF and ESDF volumetric mapping, designed for robots with RGB-D cameras. It leverages CUDA and NVIDIA hardware to build, manipulate, and query 3D reconstructions at high runtime performance.

Use cases

• reconstruct 3D scenes in real time from RGB-D camera data
• compute Euclidean signed distance fields for robot path planning
• build TSDF maps on the GPU for robotic navigation
• fuse depth camera streams into a 3D occupancy reconstruction
• query distance fields for collision avoidance on a robot
• integrate 3D scene reconstruction into a ROS2 robotics stack

When to choose

• you need real-time TSDF/ESDF reconstruction on NVIDIA GPUs
• your robot uses RGB-D cameras and needs 3D mapping for navigation
• you want a C++, Python, or ROS2 interface for volumetric reconstruction
• runtime performance of 3D reconstruction is critical

When to avoid

• you don't have an NVIDIA GPU with CUDA support
• you need CPU-only 3D reconstruction
• your project requires a permissive open-source license (license is custom/NVIDIA)
• you need reconstruction from LiDAR rather than RGB-D cameras without extra work

Facets

library · maturity active

computer-vision graphics simulation machine-learning robotics computer-vision gpu-computing simulation cpp python tsdf esdf 3d-reconstruction rgb-d cuda depth-cameras ros2 spatial-mapping gpu linux ros

Member repositories

For agents

markdown · JSON · MCP: product_card(name="nvidia-isaac/nvblox")