Image_processing

Perception step

  • Define Source,Destination points for perspective transformation
  • Apply Perspective Transform --> Warped
  • Apply thresholding
    • identify Navigable terrian
    • identify obstacles
    • identify Rock saples
  • Update Vision image (dispalyed on left side)
    • channel 0: obstacles threshold *255
    • channel 1: rock sample threshold*255
    • channel 2: nav-terrian threshold *255
  • Convert image pixels to Rover-centric coordinates for nav-terrain,obstacles,rocks
  • Convert Rover-centric Coordinates to world map coordinates
    • Rotation
    • Translation with Scale
  • convert Rover-centric Coordinates to polar coordiantes to get Angles,distances for nav-terrain,obstacles,rocks
  • update Rover world map (dispalyed on Right side)
    • channel 0: obstacles
    • channel 1: rock sample
    • channel 2: nav-terrian
  • Clear low certainty nav-terian pixels perodically(channel 2)

Decision step

can be divided into two parts
1- Dealing with normal states
2- Dealing with special states

Let's imagine a state diagram with the folling states:

- Normal states like ['forward movement', 'stop'].
- Other special states like ['sample detected', 'stuck in a loop', 'stuck by an obstacle', etc...]