Our supplementary material includes:

1. 1011-demo.mp4: 

This video (mp4 format) shows SIGNAV’s capabilities from a live demo. 
The visualization (from left to right, from top to down) includes 2D grip map, 
3D semantic point cloud, input video frames, fused depth maps, 2.5D semantic 
segmentation, and human view of the robot during the live demo. 
Note 2D grip map, 3D semantic point cloud, fused depth maps, and 2.5D semantic 
segmentation are generated and continuously updated during the live demo.

2. 1011-supp.pdf:

This PDF file includes detailed factor graph formulation for SIGNAV 
and additional 3D mapping results from Scenario 3.
