Bridging Reality to Simulation

(under construction)

CMU-Recon system builds realistic models of real-world environments by leveraging high-fidelity lidar scans and RGB images. The models can be loaded into simulation environments to support development of advanced navigation autonomy, artificial intelligence, virtual/augmented reality, etc.

To try a CMU-Recon model with our autonomous navigation systems running on the top, please refer to the 'Integrating CMU-Recon Models' section on our Ground-based Navigation Autonomy and Aerial Navigation Development Environment websites (releasing soon).

Top: A CMU-Recon model, Middle: Rendered RGB and semantic point clouds, Bottom: Rendered RGB, depth, and semantic images

The CMU-Recon model above loaded into a simulation environment, running our autonomous navigation system on the top. The rendering runs on an i7 CPU at 20Hz (no GPU involved).


Michael Kaess
CMU Robotics Institute

Jean Oh
CMU NREC & Robotics Institute

Ji Zhang
CMU NREC & Robotics Institute


Ground-based Navigation Autonomy: Leveraging system development and robot deployment for ground-based autonomy.

Aerial Navigation Development Environment: Leveraging system development and robot deployment for aerial autonomy.

Colmap-PCD: Image-to-point cloud Registration Tool.