Part of the Khronos Group
OpenGL.org

The Industry's Foundation for High Performance Graphics

from games to virtual reality, mobile phones to supercomputers

Results 1 to 8 of 8

Thread: rendering from audio/radio signal or camera view

  1. #1
    Junior Member Newbie
    Join Date
    Apr 2007
    Posts
    12

    rendering from audio/radio signal or camera view

    hi,
    well, this is not really an opengl specific question, but i hope someone here may have done this or something similar.

    PANORAMA:
    i work in a company which makes UAV's (that is, an unpiloted aircraft -- http://en.wikipedia.org/wiki/Unmanned_aerial_vehicle).

    the uav has a gps and a camera. It is wiressly transmitting data back to the pc (position, orientation, velocity, latitud, etc.. and VIDEO).

    GOALS
    Now, we have 2 objectives:
    1.-render in 3D the cameras view (that is, what the aircrafts "see")
    2.-get the coverage signal area based on the interference that could cause mountains or anthenas


    PROBLEM
    the 1st one, is our main problem and concern. The objective is that, even when we are flying on a very dark/snowy/rainny place we can see the environment we are flying over. In case that there is a mountain or bulding that wasnt planned to be there, so we can avoid it and dont crash or lose signal.

    POSSIBLE SOLUTIONS

    1.- get the geospatial data from some source like the NASA or USGS, then render the terrain info from that source and complement it by doing some simple artificial vision (creating a bitmap from the camera image).

    2.- Use something like SONAR, more specific something like ECHOLOCATION or BIOSONAR, which is the system that some animals like the bats use to "see with audio".

    3.- Use a type of radar called Synthetic aperture radar (SAR), which uses Interferometry for genereting the digital elvation model ( DEM)


    PROBLEMS WITH POSSIBLE SOLUTIONS

    1rst Solution: We wont have an internet connection all the time for getting all the data from the sites, and for having ALL the information saved, i think we would need a HUGE database. Maybe thats not a real problem, but that information would be incomplete because it cant get info from bulidings, anthenas, etc..

    2nd Solution: I think this may be the best solution, because we will be able to see "everything". The problem is that i have no idea how accurate could the image formed from an audio source will be. Also, i'm not really sure how possible is this to be done.

    3rd Solution: This may be the most accurate solution, but also the most difficult. Not only for the graphical programming part (me), but also for the electronic guys who would have to build this technology (the SAR).


    Now, based on that info, what do you recommend me?
    am i not seeing something? am i subestimating (or overestimating) the difficulty in one of he possible solutions?
    which tips, techniques, articles, etcs for implementing any of the solutions can you recommend me? (or all).
    Any new ideas?
    everything is welcomed.

    thanks in advance,

  2. #2
    Super Moderator OpenGL Lord
    Join Date
    Dec 2003
    Location
    Grenoble - France
    Posts
    5,580

    Re: rendering from audio/radio signal or camera view

    Getting 3d from vision alone is still a hard problem.
    These stereo cameras are made to build a 3D map, but no idea of the precision and the range of the depth estimates :
    http://www.ptgrey.com/products/stereo.asp

    A GPGPU lib to reconstruct from stereo or motion (looks old) :
    http://glipkit.sourceforge.net/

    Interesting comparisons here :
    http://www.gpgpu.org/cgi-bin/blosxom...VSVChallenge06

  3. #3
    Junior Member Newbie
    Join Date
    Apr 2007
    Posts
    12

    Re: rendering from audio/radio signal or camera view

    thank you.
    investigating a little bit more, i found another option:
    LASER

    Something called LIDAR may be THE solution.

    Have anyone worked with something like this? Is it a doable project for 1 person? (it can take months, if neccessary).

  4. #4
    Junior Member Newbie
    Join Date
    Apr 2007
    Posts
    12

    Re: rendering from audio/radio signal or camera view

    hey, look at this:
    http://channel9.msdn.com/posts/Rory/Micr...nd-some-genius/

    this is basically what i want to do...

  5. #5
    Senior Member OpenGL Pro Ilian Dinev's Avatar
    Join Date
    Jan 2008
    Location
    Watford, UK
    Posts
    1,290

    Re: rendering from audio/radio signal or camera view

    There was also some success in using 1 camera, with a 9-segment shutter. The shutter is a 3x3 grid of LCD (i.e only one of the 9 segments is pass-through at a time), and with software this produces a depth-map. With some nice tricks, they improved the resolution of the depth-map a lot. I can't find the url now.

    Depth-sensing cameras like that should be the way to go, imho.

  6. #6
    Junior Member Newbie
    Join Date
    Apr 2007
    Posts
    12

    Re: rendering from audio/radio signal or camera view

    well, in case someone is interested, this are my conclusions:

    LIDAR : Is the best system for doing it, but it has to much weight, so the uav cant carry it. Also it is too much expensive. Not an option.

    SONAR (Stero vision): Light weight, but in "hard" environments (rain, snow, etc..), the noise could be to much and the results wont be the expected.

    Depth Sensing camera: the range is to short.

    ISAR: good range, and "good" performance even in "hard" enviroments. the price is affortable. I had request information about the exact weight, but i still got no answer. I have read that the average weight is 30 kg. This may be the best option for the project.

  7. #7
    Super Moderator OpenGL Lord
    Join Date
    Dec 2003
    Location
    Grenoble - France
    Posts
    5,580

    Re: rendering from audio/radio signal or camera view

    LIDAR, too heavy ?
    https://mysick.com/partnerPortal/Pro...ProductID=9174
    3.5 kg - sure, it scans only a half-plane, not a full 3D cone.

    full 3D scan, at 15 kg :
    http://www.velodyne.com/lidar/produc...fications.aspx

  8. #8
    Junior Member Newbie
    Join Date
    Apr 2007
    Posts
    12

    Re: rendering from audio/radio signal or camera view

    Quote Originally Posted by ZbuffeR
    LIDAR, too heavy ?
    https://mysick.com/partnerPortal/Pro...ProductID=9174
    3.5 kg - sure, it scans only a half-plane, not a full 3D cone.
    i forgot to specify that the range needs to be >=1000 meters.
    I couldnt find a LIDAR system that has that range and weights less than 30kg (30kg is the maximiun weight supported). I Think the products you linked does not fit the needings. But thanks anyway.


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •