Weekly Report 04 (05.07. ~ 05.10.)

Overview

• LayoutNet: Reconstructing the 3D Room Layout from a Single RGB Image
• End-to-End Navigation with Branch Turning Support Using Convolutional Neural Network
• Difference of Project to Similar Projects
• Plans for Week 05

LayoutNet: Reconstructing the 3D Room Layout from a Single RGB Image

• Read through LayoutNet: Reconstructing the 3D Room Layout from a Single RGB Image
  • Summary in /docs/notes/LayoutNet.md
• Relevance for the project:
  • 3D Layout creation relevant for mapping of mobile robots surroundings
  • LayoutNet only reconstructs room layout, still need to handle "smaller" obstacles (i.e. desks, chairs, persons, etc.)

End-to-End Navigation with Branch Turning Support Using Convolutional Neural Network

• Read through End-to-End Navigation with Branch Turning Support Using Convolutional Neural Network
  • Summary in /docs/notes/EndToEndNavigation.md
• Relevance for the project:
  • project very similar to ours
    • LiDAR + front camera instead of surround view camera
    • only trajectory following, no reaction to moving obstacles

Difference of Project to Similar Projects

Feature	HCU Project	End-to-End Navigation	End-to-End Learning of Driving Models
Single Sensor System	O	X	X
Floor Segmentation / Mapping	O	O	-
Route Planning	O	O	O
Automatic Target Detection	O	X	X
Obstacle Detection	O	X	O

Plans for Week 05

1. Create first draft of mission statement (survey results)
2. Start first practical tests & attempts with floor detection and segmentation
3. Read Additional Material (optional)