Xbox Connect for 3D Cave Mapping?

Techniques, equipment and issues. Also visit the NSS Survey & Cartography Section.

Moderator: Moderators

Xbox Connect for 3D Cave Mapping?

Postby E-man » Aug 9, 2011 10:46 pm

I don't have the technical knowledge to build such a thing, but has anyone considered using an xbox connect for doing a 3D rendering of cave passages? I know people are using them for rendering interior space using a computer. Just wondering if anyone had investigated the possibility of using one for rendering cave passages. They are after all 3D scanners using infrared lasers, so I would imagine they would work just fine in a cave environment. Just thought I would throw it out there and see if anyone else had considered it or might be interested in pursuing such a project. Think about it: set the scanner down, take a 3D shot. Shoot to your next station, take another 3D shot, etc. Screw sketching, just get an accurate 3D map at every station and blend them together on the computer later to have a complete 3D map of the cave. Anyone else drooling? Yeah? I'm interested to hear other's thoughts.
-Eli
Eli Ren - 52671
User avatar
E-man
Infrequent Poster
 
Posts: 15
Joined: Sep 6, 2007 10:09 am
Location: Seattle, WA
Name: Eli Ren
NSS #: 52671
Primary Grotto Affiliation: Cascade Grotto
  

Re: Xbox Connect for 3D Cave Mapping?

Postby Bob Thrun » Aug 10, 2011 12:00 am

E-man wrote: Think about it: set the scanner down, take a 3D shot. Shoot to your next station, take another 3D shot, etc. Screw sketching, just get an accurate 3D map at every station and blend them together on the computer later to have a complete 3D map of the cave.

That is exactly what Aaron Addison did. Read the messages under the heading NSS 2011 SACS Session.
Bob Thrun
NSS Hall Of Fame Poster
 
Posts: 322
Joined: Jul 18, 2006 12:50 pm
  

Re: Xbox Connect for 3D Cave Mapping?

Postby John Lovaas » Aug 10, 2011 1:09 am

Yes, Aaron did all that- but not with an Xbox Kinect! Aaron used a very fancy(and persnickety- I think that's the technical term) terrestrial LIDAR device. Much more expensive than a Kinect. Here's an article describing the process-

http://mag.digitalpc.co.uk/fvx/ces/1104/

If someone has the interest in getting a Kinect to communicate with pre-existing software that collects and renders cave survey data in real time- such as PocketTopo- then they'd have what you want. I've seem some of the "room scanning" things the Kinect can do, and it's pretty cool. Problem is- the more 3D data you collect, the harder it is to render and use that data. We can collect boatloads of data with these devices, but it's a challenge to work with it.

It would be an awesome tool- until Microsoft releases the Kinect 2.0, with different software integration, and everyone throws their old Kinects away, and the developer who figured out how to get the 'classic' Kinect to collect cave survey data doesn't know anything about the new programming language the Kinect 2.0 uses... ;-)
imbecile sheepherder.
User avatar
John Lovaas
NSS Hall Of Fame Poster
 
Posts: 590
Joined: Sep 6, 2005 9:10 am
Location: Woodstock, Illinois
  

Re: Xbox Connect for 3D Cave Mapping?

Postby Aaron Addison » Aug 12, 2011 4:50 pm

The short answer is "yes". The main issue with Kinect is that the RGB-D sensor in the camera is only good out to about 8m. They figured that was about as far away from the TV a gamer would be. There are companies currently working on an industrial version of the Kinect camera that would work at longer distances. The guys at Washington University (in Washington State) have done it already.

http://www.youtube.com/watch?v=MRn34pmD5ww

They even wrote a simple loop closure routine that adjusts based on the point cloud. If you watch the video closely, you can see it in action. Here is another link that shows the interaction with the point cloud collected in Kinect.

http://www.youtube.com/user/RGBDvision# ... u5Ywwb4RaU

Microsoft just released the Kinect SDK a month or so ago, So I would expect to see much more development.

http://research.microsoft.com/en-us/um/ ... kinectsdk/

If anyone is motivated to write some code specific to caving, I'll send you a Kinect.

Cheers,

Aaron
------------------------------
Aaron Addison
Stop and ask yourself - "What is this map for?"
Aaron Addison
Prolific Poster
 
Posts: 180
Joined: Apr 11, 2006 8:03 am
Location: St. Charles, MO
  

Re: Xbox Connect for 3D Cave Mapping?

Postby KeyserSoze » Aug 15, 2011 11:24 am

Great links Aaron. I've been watching other "kinect hacks" videos and I've been really impressed with how many unique developments have already been made by amateurs. I've said before that 3d technology like this will be the future of cave mapping, I just didn't realize how close it was to becoming affordable and practical.

Check out the 3d scanning equipment on this web-article, "the Nottingham Caves Survey crew hauls equipment below the surface on bike trailers."
http://webecoist.com/2011/01/17/mapping-the-underworld-digital-3d-cave-exploration/

With the kinect you could go into a cave with the little black bar and a laptop in a backpack to run the software. You wouldn't have a perfect 3d map like the ones in the article but you could certainly gather enough data to draft a typical plan/profile view cave map while in the comfort of home. Think of how much time and manpower it would save.
This signature is really funny
User avatar
KeyserSoze
NSS Hall Of Fame Poster
 
Posts: 227
Joined: Nov 6, 2007 2:18 pm
Location: Louisville, Kentucky
NSS #: 61069
Primary Grotto Affiliation: Louisville Grotto
  

Re: Xbox Connect for 3D Cave Mapping?

Postby rlboyce » Aug 15, 2011 1:27 pm

Honestly, this technology could be applied to cave mapping RIGHT NOW. The problem is getting someone who knows how to deal with the software, and perhaps more importantly, sufficient lighting. I guess this wouldn't be ideal for a very wet cave either.
Happiness can be measured in mud/in².
User avatar
rlboyce
NSS Hall Of Fame Poster
 
Posts: 727
Joined: Jan 22, 2010 7:26 pm
Location: Chambersburg, Pennsylvania
Name: Ryan Boyce
  

Re: Xbox Connect for 3D Cave Mapping?

Postby William Tucker » Aug 15, 2011 2:11 pm

The Kinect will work in total darkness. It works by projecting a grid of infrared dots on the scene then photographing it. Trigonometry takes care of the rest. The issue has been trying to stitch multiple frames together into a single model but there are those who are working on it. Google "KinectFusion" or take a look at this YouTube video: http://www.youtube.com/watch?v=quGhaggn3cQ.
User avatar
William Tucker
Frequent Poster
 
Posts: 52
Joined: Mar 7, 2007 7:03 pm
Location: Wolfforth, TX
Name: William Tucker
NSS #: 55566RL
Primary Grotto Affiliation: Lubbock Area Grotto
  

Re: Xbox Connect for 3D Cave Mapping?

Postby Aaron Addison » Aug 15, 2011 4:53 pm

Correct about Kinect working in the dark. The only bummer is that since you are working with IR instead of reflectance (LiDAR), the point cloud is monotone.

Of course this is a small price to pay for the cost difference.

AA
------------------------------
Aaron Addison
Stop and ask yourself - "What is this map for?"
Aaron Addison
Prolific Poster
 
Posts: 180
Joined: Apr 11, 2006 8:03 am
Location: St. Charles, MO
  

Re: Xbox Connect for 3D Cave Mapping?

Postby rlboyce » Aug 15, 2011 5:25 pm

William Tucker wrote:The Kinect will work in total darkness.


In that case... WTF are cavers waiting for??? Sure, the technology is not perfected for cave use yet, but IMO it's certainly far enough along now to give it a shot to see what happens.
Happiness can be measured in mud/in².
User avatar
rlboyce
NSS Hall Of Fame Poster
 
Posts: 727
Joined: Jan 22, 2010 7:26 pm
Location: Chambersburg, Pennsylvania
Name: Ryan Boyce
  

Re: Xbox Connect for 3D Cave Mapping?

Postby William Tucker » Aug 15, 2011 6:30 pm

Who says we are waiting? I have been working on this for months now; even before the SDK release. It is happening and fast. It is not hard to predict the demise of traditional survey, sketching and drafting techniques. For volumetric mapping, you don't need the RGB camera hence no light is needed either. If you want to go the next step and wrap the volume with textures derived from the RGB camera, all you need is light. Traditional mapping techniques are going to be replaced and the Kinect is a good step in that direction. It doesn't replace inventory data but it does a far better job at mapping with much more detail than can be captured in a traditional sketch or map. And, there is no need to place survey stations (except for inventory purposes) as the device effectively places thousands of virtual survey stations as it views the space.
User avatar
William Tucker
Frequent Poster
 
Posts: 52
Joined: Mar 7, 2007 7:03 pm
Location: Wolfforth, TX
Name: William Tucker
NSS #: 55566RL
Primary Grotto Affiliation: Lubbock Area Grotto
  

Re: Xbox Connect for 3D Cave Mapping?

Postby rlboyce » Aug 15, 2011 6:46 pm

I assumed no one has been actively pursuing this form of mapping as no one has been talking about it.

How far along are you on your attempts?? Surely you've tried it in a cave by now, and if so, do you have any examples of snapshots of a model to share with us? If not, I suppose I'm rather confused as to what the issue really is... seems to work quite painlessly in the videos.
Happiness can be measured in mud/in².
User avatar
rlboyce
NSS Hall Of Fame Poster
 
Posts: 727
Joined: Jan 22, 2010 7:26 pm
Location: Chambersburg, Pennsylvania
Name: Ryan Boyce
  

Re: Xbox Connect for 3D Cave Mapping?

Postby William Tucker » Aug 15, 2011 7:14 pm

Getting a 3D model of the space immediately in front of the device is relatively easy. Moving it around and stitching the various frames together into a larger model is much harder (at least for me). I am working on the software side and have not had a chance to try anything in cave; but, it works just fine in my house.
User avatar
William Tucker
Frequent Poster
 
Posts: 52
Joined: Mar 7, 2007 7:03 pm
Location: Wolfforth, TX
Name: William Tucker
NSS #: 55566RL
Primary Grotto Affiliation: Lubbock Area Grotto
  

Re: Xbox Connect for 3D Cave Mapping?

Postby rlboyce » Aug 15, 2011 7:25 pm

Sweet. I am envious that I do not have the knowledge to do what you are doing. Feel free to give us updates in the future! I wish you luck on your project.
Happiness can be measured in mud/in².
User avatar
rlboyce
NSS Hall Of Fame Poster
 
Posts: 727
Joined: Jan 22, 2010 7:26 pm
Location: Chambersburg, Pennsylvania
Name: Ryan Boyce
  

Re: Xbox Connect for 3D Cave Mapping?

Postby Aaron Addison » Aug 15, 2011 7:40 pm

yes, we have been working on kinect almost since the time it was released. The crux move (to steal a rock climbing term) is to stitch the static data clouds together on the fly. Even now though it is not difficult to use traditional survey techniques, then take a Kinect scan at each station. Various software can be used to then "reassemble" the clouds to form a 3D model of the cave. ALL of this can be done for less than $500. Is the technology v1.0 beta? Absolutely. But as others have already observed, it is coming....and fast.

Look towards the gaming community to drive the development. There are MANY more gamers out there than cavers (let's hope it stays that way). They want to be able to load real world terrain in to their gaming engines for massive multi-player online games. The easiest path for them looks to be the Kinect (or some variation).

From the day I first saw a Kinect, it changed what I believe to be the future of cave survey. It is a disruptive technology that will likely change almost every aspect of how we document caves.

Cheers,

AA
------------------------------
Aaron Addison
Stop and ask yourself - "What is this map for?"
Aaron Addison
Prolific Poster
 
Posts: 180
Joined: Apr 11, 2006 8:03 am
Location: St. Charles, MO
  

Re: Xbox Connect for 3D Cave Mapping?

Postby William Tucker » Aug 16, 2011 8:01 am

O.K. Let me see if I can describe the problem as I see it to give you some idea. Maybe this will generate some thoughts from others that might help in solving it.

Given a point cloud (a set of X,Y,Z coordinates) as the model and a new point cloud (frame) which may need to be joined to the model, the new frame could be (and most likely is) distorted in a number of ways. First, it could be shifted along the X, Y or Z axis of the viewer or some combination of all three. Next, it could be rotated around the X, Y or Z axis of the viewer or some combination of all of these. Remember, it is the viewer that is moving, not the scene, usually. That fact cost me some effort as I was going down the wrong path at first. And, both shifts and rotations can and do occur simultaneously. Also, the new point cloud does not have to align point by point as it may (and usually does) align with the spaces between the model points. The problem is (as I see it) to calculate a transformation matrix to apply to the second frame to join it to the model. This transform is the start of the transform for the next and subsequent frames, too. New data can show up from the second frame. This new data could be more or less detailed and it does contain some amount of noise or there may be noise in your model which needs to be removed. The worst problems are when the new frame is distorted from the first by both time and space (in other words, it does not join or the distortion is severe enough that you cannot identify how it was distorted). Combine that with changes in the scene such as someone walking into frame and stitching is a tough problem. I am not convinced that anyone is doing this in real-time, yet.

And, I am almost certain that I am doing this the hard way. My background is not in graphics nor noisy instrument data. It looks to me like some may be using the GPU which I know very little about.

As to inventory, I have thought of the idea of aiming the device at an item which needs to be inventoried and pressing a button to generate a new inventory station code which is used to record the artifact, mineral, fossil or whatever and locate it within the model.
User avatar
William Tucker
Frequent Poster
 
Posts: 52
Joined: Mar 7, 2007 7:03 pm
Location: Wolfforth, TX
Name: William Tucker
NSS #: 55566RL
Primary Grotto Affiliation: Lubbock Area Grotto
  

Next

Return to Survey and Cartography Forum

Who is online

Users browsing this forum: No registered users

cron