princeton-vl/DROID-SLAM

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2671 stars · 427 forks observed · 2026-08-28

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

DROID-SLAM is a deep learning-based visual SLAM system for monocular, stereo, and RGB-D cameras that estimates camera trajectory and dense 3D reconstruction. It is a PyTorch/CUDA research library from Princeton Vision Lab requiring a GPU for inference.

Use cases

• estimate camera trajectory from monocular video
• run visual SLAM with stereo or RGB-D cameras
• reconstruct dense 3D depth maps from video
• benchmark SLAM on standard datasets
• run SLAM inference on multiple GPUs

When to choose

• you need deep-learning-based SLAM with strong accuracy on monocular, stereo, or RGB-D input
• you have a CUDA GPU with at least 11GB memory and a PyTorch workflow
• you want dense depth reconstructions alongside camera poses

When to avoid

• you need lightweight real-time SLAM on CPU or embedded hardware without a GPU
• you need a production-ready plug-and-play SLAM solution rather than research code
• you lack CUDA-compatible hardware

Facets

library · maturity active

computer-vision machine-learning deep-learning simulation computer-vision robotics autonomous-vehicles deep-learning python cpp slam visual-odometry monocular-depth stereo rgbd pytorch camera-pose-estimation 3d-reconstruction gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="princeton-vl/DROID-SLAM")