Bathymetry for next release playground : Candiate 2. Santorini

[woensug-choi]

Santorini

Launch command

roslaunch dave_demo_launch dave_Santorini_demo.launch
python3 dave/examples/dave_nodes/src/merry_go_round.py

Merry-go-round script GIF Animation (Click the image to see the animated GIF)

New files in this PR

• world file : dave_Santorini.world
• launch file : dave_Santorini_demo.launch
• Model : dave/models/dave_object_models/models/Santorini
• Merry-go-round-script dave/examples/dave_nodes/src/merry-go-round.py

Statistics

• Origin of the source data : EMODnet Bathymetry Consortium (2020): EMODnet Digital Bathymetry (DTM).
  https://sextant.ifremer.fr/record/bb6a87dd-e579-4036-abe1-e649cea9881a/
• Resolution : 75 cm
• Latitude : 36.337 ~ 36.479
• Longitude : 25.317 ~ 25.489
• Size of the patch : (Lat, Lon) = (0.015, 0.014) = Approx 1000 x 1000 meters)
• Maximum Height : 554.95 m (land)
• Minimum height : -453.93 m
• Number of vertices of original data = 350 x 291

Conversion method

• mkbaty.py automation script at https://github.com/Field-Robotics-Lab/Bathymetry_Converter
• Unlike the axial seamount case, it needed some manual manipulation of the source data
  • because the XYZ in the GeoTiff was not equally-spaced. It had to be interpolated using gdal_grid
• Flags parameters used inside the script :
  • ONE_TILE_AT_ORIGIN : True
  • SIMPLIFICATION_LEVEL : 0
    • Size of the OBJ file is 18.9 MB
• Parameters used:

  • PREFIX = Santorini

  • Ranges

    • STARTLON = 25.317
    • STARTLAT = 36.337
    • ENDLON = 25.489
    • ENDLAT = 36.479

  • Color texture (color.txt)

    100 213 62 79
    50 244 109 67
    0 254 224 139
    -50 255 255 191
    -150 230 245 152
    -200 171 221 164
    -250 102 194 165
    -300 50 136 189
    -350 94 79 162
    -400 77 77 77
    -450 26 26 26
    

Notes

• Z-buffer in the image is very shallow. @WilliamLewww Do you know the correct settings for the camera at the end of the world fine to increase the visible area?

• As mentioned by @bsb808, it includes complex geometry in the small region to make bathymetry interesting. But no vast planar regions. I have a hunch that we would choose this instead of the axial seamount because we all want to visit even in the simulation.

• It would be simple to place multiple mission tasks at different angle directions from the center and modify the merry-go-round script to relocate for each task to demonstrate. (0 degrees location = sonar local search , 120 degree location : manipulation, 240 degree location : other tasks)

[WilliamLewww]

You might want to increase the far clip plane. It already seems to be a large value, but increasing it further would help with worlds that are larger.

[woensug-choi]

What we see in the gif is as far as it can get. Or something is wrong in the world file nothing much changes...

[WilliamLewww]

Hm, it might be a limitation in OGRE. I'll go check to see if there is a parameter we could set to change the view distance.

[WilliamLewww]

The artifact is caused by the terrain and not the camera. When I hard-code the camera in gazebo to have an infinite near-far clip plane, I still get the same amount of visibility. I am currently investigating the terrain visual.

[woensug-choi]

Thanks! Hope we figure out something. If it is a limitation, it's fine for now :)

[mabelzhang]

Verbal discussions from meeting:
Increase camera angle and still doesn't solve issue.
Per @WilliamLewww: Deep issue in rendering. Tried clipping at far = 0.
Related bug per @woensug-choi: No matter the xyz relative to the object, initial state still very close, very zoomed in.
Per @M1chaelM: A possibly related issue has been ticketed in osrf/vrx#336

[j-herman]

Everything runs fine on my machine and now I want to go to Santorini! @woensug-choi , do you want to play more with the depth on the rendering, or do you think this is all we can get? It looks good when you are zoomed in on the vehicle as we would be for each demo activity, but it is a little confusing to watch from farther out.

[woensug-choi]

@j-herman I guess people would like this region better :) I would like to wait for @WilliamLewww to figure out about visibility improvement and #194 to merge.

The next steps would be, to decide which one to merge from two candidates, invite others to put their stuff on the scene (for me, I would add models from #194 and add multibeam sonar to Rexrov to demonstrate a multibeam sonar local search example at this bathymetry), and generate a script that would relocate the Rexrov for each activity area.

It feels great to push things forward day and night taking the advantage of the time differences :D

[Yadunund]

Can we merge this in? We'd like to use the Santorini model as the basis for the integrated world.

[woensug-choi]

Merging it without @j-herman 's review approval. She tested it but just did not click the approval button.

[woensug-choi]

Documented at https://github.com/Field-Robotics-Lab/dave/wiki/Bathymetry-Models