Multi-view 3D reconstruction and mapping (showreel)
Dense mapping, nvblox-style fusion, and Isaac task environments for navigation and manipulation.
Capabilities
This compilation spans dense mapping and scene understanding from multiple cameras, volumetric and surfel-style mapping (including nvblox-style pipelines), and fusion of depth and color for both navigation and manipulation. Work here also ties into simulation arenas such as Isaac Lab and kitchen-style task environments so reconstructed scenes and synthetic scenes can support the same downstream stacks.
How we demonstrate it
We show the work through side-by-side clips and stills from several engagements: multi-camera reconstruction runs, mapping visualizations, Isaac and Omniverse environment shots, and exports such as point clouds or generated assets that feed perception and motion planning. Together they document how we go from sensors to maps to robot-ready representations.
Highlights
Together these artifacts document a repeatable path from sensors to maps to robot-ready scene representations for navigation and manipulation stacks.
Technologies
Gallery
Project Media
