Shop OBEX P1 Docs P2 Docs Learn Events
Visualizing 3D Interactive Map — Parallax Forums

Visualizing 3D Interactive Map

rough_woodrough_wood Posts: 61
edited 2009-12-06 07:20 in Propeller 1
I am interested in making a robot that can build a 3D map. For example it could use a Ping on a Pan-Tilt.

My issue, is I would like to be able to create a map that can be sent to a user who can view the map, and do simple things like rotate/view the map, control the bot, set waypoints relative to walls/object. Probably using a laptop, or even more fun, a screen and home brew controls.

I am wondering how helpful Hydra would be for this application? I prefer to do it in a cheaper way if possible, but from my novice view it seems the Hydra could be a good way to do it.

What do you guys think the best way to do this would be? I need to order this week so I can work on it during break, while I have access to gear I have access to up north. I have several Exams so I don't have as much time for Google as I'd like before I need to order. Plus without having used Hydra it's hard for me to know what capability it has in this application.

I should add I have access to a good student version of Matlab which I am sure I could get to work well. But I would still like to hear your input with regards to Hydra, or if it can be done much cheaper with just another Prop. I want to do as much of this outside of Matlab as possible. This is more of a learning project than a real world thing I intend to sell.

The main issue is user interface, and not the Bot itself. Again, this is a learning project and not a real world product. I want to create as much as I can with my limited knowledge and avoid even controlling with Matlab-Laptop if possible (but without incurring a large $$ hit, I do consider $200 a somewhat large hit)

I am sure you will need more information before you can give me any good answers, but I am looking forward to your guys advice. You have way more knowledge about this type of thing than I do, especially Hydra I predict. I am a Junior getting BS in ME, to give you an idea of my background.

Thanks for the help!

Keith

Comments

  • rough_woodrough_wood Posts: 61
    edited 2009-12-05 22:35
    I am not finding any true 3D object rotating successfully done with the Hydra, has no one been able to do it?

    It looks like I'll have to do the user interaction with Matlab, unless there is a way to do it with Hydra or Prop that I am unable to find.

    For that matter I'm not sure a map or "level" can be made on the fly with Hydra either. It looks like it pretty much is dedicated to oldschool Atari type games.

    I will begrudgingly begin with Matlab unless I hear good news from someone here.
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2009-12-06 00:32
    The propeller on a hydra board or otherwise can do a lot but you are a bit vague about what you are expecting. You can certainly do wire frame 3D on the prop or low resolution point clouds and maps are just a form of 2D graphics but it does depend on what you need or expect. Beau did a 3D demo someone might have a link to.

    I would have thought a proto-board would make more sense for a robot project than the hydra or one of the robot specific boards.

    Graham
  • rough_woodrough_wood Posts: 61
    edited 2009-12-06 00:46
    I would like to get shaded faces, which would probably be aproximated by planes covering the point clouds. Objects could be aproximated by flat and spherical surfaces for example, mostly to show the region in which the bot could move.

    Yes I would have a Prop in the bot, but the Hydra, as I was picturing I might use it, would be on the user end outside the rubble pile or small cave etc.

    I have about an hour I can spend looking tonight, but it's looking like I'll have to do it in Matlab. Comprehensive Calc III Final followed by 3 more, 2 of which are also comprehensive [noparse]:([/noparse].

    Interested in seeing that link though! I'm hoping to do as much as I can without it though. I "know" Matlab, the goal is struggle through something I don't. Even if the finished project isn't as tangibly useful after I'm done making it.

    Thanks for the response.
  • potatoheadpotatohead Posts: 10,261
    edited 2009-12-06 00:48
    Seems to me a single board computer, running Linux and OpenGL would be a great target for this.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Propeller Wiki: Share the coolness!
    Chat in real time with other Propellerheads on IRC #propeller @ freenode.net
    Safety Tip: Life is as good as YOU think it is!
  • rough_woodrough_wood Posts: 61
    edited 2009-12-06 00:50
    I should add, wireframe may be adequate if gridlines can be drawn to make spatial awareness better for the user. It pretty much has to be rotatable though, the 3D shell that is. This is even more important the less clearly the image is represented. Point clouds would need to rotate cleanly and quickly since it is so abstract.

    A pixel is worth a thousand bits, a video of pixels moving showing shape is worth a thousand bytes.
  • rough_woodrough_wood Posts: 61
    edited 2009-12-06 00:55
    @Potatohead

    I am looking for a Ubuntu laptop as we speak. I got an Asus but they put in the wrong hardrive, so I either send it in as DOA, or swap it for a Dell Mini10v or an Asus 1000HA-P.

    I don't know the first thing about Linux however. I tried installing Ubuntu from "Live CD" on a USB Flash, but it fails at 82% because of some network failure I need to fix with a kill command I think. Couldn't get around it and ran out of time to do it before I leave for Michigan.
  • SciNemoSciNemo Posts: 91
    edited 2009-12-06 01:22
    Wait, don't go for that laptop yet!

    I have been working on this problem for a time, and here is the method I have come up with:

    Setup:
    -Pan/Tilt servo head
    -Ping sensor mounted on pan/tilt
    -(2)Propeller Protoboard
    -(2)Serial transmitter/receiver
    -Wall wart (for power)

    Method:
    1. Propeller uses pan/tilt to scan in circles, 360 degrees to a circle. At the end of each circle the tilt of the ping sensor is moved 1 degree upward. A measurement is taken every degree.
    2. When all the measurements are taken the data that has been collected (distance and degree of x/y rotation) is transmitted to the laptop that the user has.
    3. Program on laptop constructs 3d map from data sent from Prop robot.
    4. Path finding and display of map is done using laptop program.
    5. Path data is deconstructed into simple forward, backward, turn left, turn right, etc. instructions for the robot.
    6. Simple instructions are sent back to robot.
    7. Robot follows the simple instructions.
    8. When all instructions have been followed, goes to step 1.

    A note: However you do this, if you are using a simple pan/tilt and ping sensor it will be extremely slow. A single scan will most likely take several minutes.

    Also, the map will be very inaccurate. Acoustic ping sensors do not like soft/angled/echoey objects, which return unexpected results. I've tried it with lasers, but the resolution isn't as good as I would like. To solve this, you could buy an electronic laser tape measure off of ebay and interface it to a microcontroller (propeller) somehow, but I haven't gotten that far yet.

    Whatever approach you take, good luck with the project, and I expect to see pictures at some point smilewinkgrin.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Not the fish.
    sites.google.com/site/bitwinproject/
  • rough_woodrough_wood Posts: 61
    edited 2009-12-06 01:48
    That is more or less what I plan to do, SciNemo.

    I wasn't planning on taking quite so many data points as your are. I mainly want a system that works for when I get good sensors, I will have a good platform to start from.

    Why do you say "Wait, don't go for that laptop yet!"? Something wrong with it for the application? Something better you recommend? I have a few days before I need to decide on the Laptop.

    I basically want simple hardware, and intend to continue to tweak the software. Such as having it measure again, and if an object seems to have moved, maybe watch it and try to calculate speed. Set a waypoint and have the bot get there by watching it's distance according to landmarks. Things of that nature. Obviously all that won't happen in a week. Just a side project to take up some time for a while.

    Edit: Also, do you say stop due to a software compatibility issue, or a hardware specs issue? I also have a XP laptop which runs Matlab fine. I want a very portable laptop under say 3Lbs for this project. 10" is about perfect, I think.

    Thanks

    Post Edited (rough_wood) : 12/6/2009 1:54:35 AM GMT
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2009-12-06 01:55
    This is the link:

    http://forums.parallax.com/showthread.php?p=605385

    But to be honest I think what suggest you want is probably somewhat beyond what you might expect to do with the propeller. I'm not so sure it would be all that easy in matlab either actually.

    I'd probably start by getting the robot and the scanning working first and let that steer your course with regards to visualisation and interaction.

    Graham
  • rough_woodrough_wood Posts: 61
    edited 2009-12-06 07:20
    Thanks for the link.

    Yeah looking at the demo it looks like you pretty much need your points and lines pre-determined. I don't think the assembly can just be changed while the prop is running in that Dat section. I'd have to set up some system to throw data points to memory address' then have the program somehow know what lines to connect to what points.

    Basically it looks like I was right that I shouldn't use the money on Hydra right now.

    I'll take a better look at Matlab. make sure my little netbook will have min requirements or above.
Sign In or Register to comment.