It is currently Fri Oct 20, 2017 5:21 pm



Reply to topic  [ 184 posts ]  Go to page 1, 2, 3, 4, 5  Next
 Red Rovr Motion System (formerly Friispace) 
Author Message
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
A while back there was a long thread about how to build an omnidirectional motion device. A lot of ideas got kicked around and one of them was just creating a DIY free-motion detection system using Wiimotes and IR lights. I championed the idea, but it didn't seem to generate much interest since it didn't exactly fall into the category of a compact indoor device.
http://www.mtbs3d.com/phpBB/viewtopic.php?f=120&t=13784&start=195#p65785

The idea stuck with me and I've been tinkering with it over the last few weeks and have finally gotten it to the point where there is something to show. Right now it's still in early prototype phase with crude software and small scale mockups but soon I should be able to scale it up to human size. Here's a video of where I'm at right now...



The basic idea is that you could go out to a basketball court, mount a Wiimote camera (or 2) on top of the goals, calibrate the system, and using an HMD+head tracker+backtop be able to control player motions in a game using real-life unencumbered movements. Simple, cheap, and highly realistic! The downsides are of course that it really only works well in a large space and also that there is a physical boundary problem.

What happens when you reach the edge of your play-space? Currently my idea is to automatically pause the game and dim the screen when the player moves outside of the play area. The player must then turn around and step back into bounds for the game to automatically un-pause. Admittedly a crude and jarring solution but hopefully with practice the player would learn how to perform the maneuver quickly and only cause "hiccups" of a second or two. Additionally the player can "scale" the field so that small movements cover larger distances in game space - so instead of 1:1 walking speed you could have shuffle walks translated into normal game speed walking. This would effectively extend your playing field. I'm also planning on researching much larger spaces (football size) using similar techniques. That opens up the possibility for using scaled rotation and perceptual tricks to guide the player in circles even though they think they are moving in straight lines within the game.

I've got a lot of work to do yet. Next up is integration of the head tracker. Then I've got to work on scaling up to larger sizes. Multiple cameras, multiple light sources, etc.... I've got dozens of ideas and different variations on this theme to sort out and I probably won't have anything playable for a good while. I'm going to keep this as a project diary and idea space and I'll try to post my progress as I meet important milestones.


Last edited by brantlew on Sat Sep 29, 2012 9:52 pm, edited 5 times in total.



Sat Dec 03, 2011 12:47 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Jan 09, 2010 2:06 pm
Posts: 2255
Location: Perpignan, France
Reply with quote
Nice that you got this started.

For your idea of scaling the field to not have problems with the edges of the play-space, you can have a look at "redirected walking". There's been quite a lot of ongoing research on this subject.

An introduction :
- Redirected Walking – Playing with your perception’s limits.

Some publications :
- Redirected Walking ;
- Generalised Algorithms for Redirected Walking in Virtual Environments ;
- Estimation of Detection Thresholds for Redirected Walking Techniques.

Video : Using Avatars for Redirected Walking
Related paper : Velocity-dependent curvature gain and avatar use for Redirected Walking


Good luck with your project !


Sat Dec 03, 2011 2:12 am
Profile WWW
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
@Fredz: Thanks for the links. That's exactly the type of thing I had in mind. I just have no idea how much space is actually required to imperceptibly lead the player in a circle. They did it in a 9x9m area which is impressive. I would love to be able to get it to work in my "basketball" size playing field but I assumed that it was too small. Since I don't have access to the in-game data (this is only an i/o peripheral), its a little harder because I can't implant obstacles or distractions into the game, but I still think it may be possible if the player turns their head often which is common in an FPS. This is one aspect of the project that I am the most interested in.


Sat Dec 03, 2011 8:18 am
Profile
3D Angel Eyes (Moderator)
User avatar

Joined: Sat Apr 12, 2008 8:18 pm
Posts: 11364
Reply with quote
Very nice work. I do wonder though if the Wii camera can track all the way over a basketball court. I guess it could with a powerful IR light.

_________________
check my blog - cybereality.com


Sat Dec 03, 2011 10:39 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Jan 09, 2010 2:06 pm
Posts: 2255
Location: Perpignan, France
Reply with quote
In the first link they said this :
"For example, we know that we can guide them on a circle with a radius of approximately 23 meters, while they believe that they walk straight in the VE. This guidance approach can be realized by injecting small rotations to one side, which enforce users to unknowingly walk on a circular arc in the opposite direction."

I guess it must be dependant on the walking speed and probably some other factors though. It also doesn't account for direction changes like in a real environment, but that still gives a good idea of the limits of the technique.


Sat Dec 03, 2011 10:50 am
Profile WWW
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
cybereality wrote:
Very nice work. I do wonder though if the Wii camera can track all the way over a basketball court. I guess it could with a powerful IR light.


I bought a pack of 20 IR LEDs. I'm going to wire 5 or 10 of them and shove them in a ping-pong ball. Hopefully that will be bright enough to track over that distance.


Sat Dec 03, 2011 11:23 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
Fredz wrote:
In the first link they said this :
"For example, we know that we can guide them on a circle with a radius of approximately 23 meters, while they believe that they walk straight in the VE. This guidance approach can be realized by injecting small rotations to one side, which enforce users to unknowingly walk on a circular arc in the opposite direction."

I guess it must be dependant on the walking speed and probably some other factors though. It also doesn't account for direction changes like in a real environment, but that still gives a good idea of the limits of the technique.


I haven't read the paper yet but that is an interesting quote. A 46 meter diameter circle is a pretty big space (about half a football field). Hopefully that is the worst-case scenario - where the player is looking straight ahead and walking forward without stopping. My intuition tells me that if the player turns their head a lot or makes plenty of turns then you can take advantage of that and guide them into a more aggressive radius.

Edit: Just read the article. A 46 meter diameter space is a good guideline for worst-case straight line walking. That fits just about perfectly into half of an American football field and makes for a nice "easy-to-state" project end-goal. If all the equipment ranges and sensitivities can be scaled to half a football field, then (in principle) you could provide a totally unbounded free walking game experience.


Sat Dec 03, 2011 12:11 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Aug 21, 2009 9:06 pm
Posts: 1644
Reply with quote
Wow, that is really impressive!

A few hardware suggestions:

1) The Wiimote has a very low field of view, it might be hard to get it far away/high enough to cover the whole court. Would using another camera be possible, or are you relying on the onboard processing of the Wiimote? If so, you might be able to add a wide angle lens, though that would probably greatly complicate things. The Wiimote camera is low res, too, so for really large spaces it might not give sufficient detail. Multiple wiimotes could solve that, I imagine.

2) You will probably want to sand down the front face of the LEDs. It acts as a focusing lens, and for this kind of application, you want a diffuse light, not a focused one.

If you need anyone to help you test, I would love to do so!

One more thing: How is the game receiving this input? Does it see it as keypresses, or an analog gamepad, or something else?


Sun Dec 04, 2011 12:48 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
@Palmertech: Thanks! I would love to combine this tech with your PRx HMD's to create a low cost version of that million dollar VR room for something like $2000!

As to your comments - I'm not terribly worried about the resolution of the Wiimotes right now. Since I am only tracking gross character motion I don't really need a lot of precision. I think a few centimeters per pixel is adequate to accomplish what I'm trying to do. In some ways the low resolution may help by acting as a sort of natural filter to remove all the extraneous motion like head bobbing and head turning out of the signal. Camera coverage is an issue though and I'm thinking of using multiple cameras to handle that. For the basketball court scenario here are 3 possibilities with 2 cameras that I am considering...

Image

The green bar shows the area of camera coverage. The first pattern would offer a higher resolution. The second one would offer a larger surface area. The third would be useful for triangulation. Right now I cannot determine height changes with only a single camera so ducking has to be detected with an inertial sensor. Overlapped cameras could be used for height tracking. Additionally they could be used to track other objects in 3D (ie. gun, sword, tennis racket, etc) that might prove useful.

Edit: Right now I'm bridging input from my software to the game via GlovePie and using key commands exclusively. At some point I will probably just implement direct keyboard/mouse emulation and bypass GlovePie, but for now it's useful because the game I am testing with (theHunter) supports TrackIR which GlovePie can emulate.


Sun Dec 04, 2011 5:22 pm
Profile
Certif-Eyable!
User avatar

Joined: Sat Dec 22, 2007 3:38 am
Posts: 990
Reply with quote
from what I'm seeing, using a single IR LED, you are assuming constant height for the person. If they duck, then it would read this as a move towards the camera.
I wonder if IR sneakers with LED on the front and back and sides could work better. Have a switch in the sole so that they are only tracked when touching the ground. Using some smoothing and motion prediction (at the point that both shoes are on, you can assume the person is half way between the two points, you can probably assume they are moving the same speed as they were moving between the last two times both shoes were on) and you should be able to get some nice tracking this way. For running, you will never have both on at the same time, so you would need to monitor for the switching from one to another.


Sun Dec 04, 2011 6:03 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Aug 21, 2009 9:06 pm
Posts: 1644
Reply with quote
When I have enough time to build a few more PRx systems, I will definitely send you one to try out with this. The idea of going out at night to play around with VR stuff on the local basketball court sounds awesome!

EDIT: Not exactly sure how you would mount it, but it seems like one camera on each end of the court would be great. One solution that seems elegant would be to make a metal/plastic ball that is slightly larger than a basketball, with the camera mounted inside, facing out. Just get on a small stepladder, place it so that it is resting on the rim of the hoop, and boom! Super stable, fully adjustable system that can be used in both indoor and outdoor courts.


Sun Dec 04, 2011 7:33 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
android78 wrote:
from what I'm seeing, using a single IR LED, you are assuming constant height for the person. If they duck, then it would read this as a move towards the camera.
I wonder if IR sneakers with LED on the front and back and sides could work better.


Yes, with a single camera ducking would register as movement towards the camera. The sneakers idea sort of scares me - not saying it couldn't work but it seems like a really noisy signal to work with. It does have one advantage though - it puts an extra 5 or 6 feet of height difference between the light and the camera and extra height difference makes the coordinate calculations more accurate. Right now however, I am just planning on trying to handle height with inertial sensors - so hopefully I can adjust the optical coordinates with the inertial readings.


Sun Dec 04, 2011 9:49 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
PalmerTech wrote:
When I have enough time to build a few more PRx systems, I will definitely send you one to try out with this.


Excellent! I was hoping you would say that. I'm definitely up for a technology swap. I've been dying to check out one of those units. :D

PalmerTech wrote:
EDIT: Not exactly sure how you would mount it, but it seems like one camera on each end of the court would be great. One solution that seems elegant would be to make a metal/plastic ball that is slightly larger than a basketball, with the camera mounted inside, facing out. Just get on a small stepladder, place it so that it is resting on the rim of the hoop, and boom! Super stable, fully adjustable system that can be used in both indoor and outdoor courts.


That's really a great idea. It mainly depends on how much height I need on the cameras. I'm skeptical about just 10' which is why I'm planning on mounting at the top of the backboard (and maybe higher) but it would be great if just 10' was sufficient. If you look in the video I am currently using one of those twisty-claw-camera-mount-thingies which can mount just about anywhere.


Sun Dec 04, 2011 9:57 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Aug 21, 2009 9:06 pm
Posts: 1644
Reply with quote
If 10' is not enough, then you could mount a rod on top of it. Hmm... I wonder if you could rig something up that has a collapsible pole going to some kind of clamp, then another collapsible rod going up out of the clamp, and attach the camera to that. The idea would be that you extend the pole on both ends, and clip it onto the backboard or something without any need for a ladder.


Sun Dec 04, 2011 10:12 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
@PalmerTech: I like how you're thinking... I've also been looking into 25' telescoping flagpoles that have an L-shaped base that you stabilize with a vehicle tire. But the basketball court offers a very unique and fortuitous situation because it already has stable mounting points at perfect locations. There seem to be a lot of possibilities.

Football fields also have nice standardized mounting points at the goalposts (and conveniently labeled measurements ;) ) but they are not nearly as convenient or accessible.


Sun Dec 04, 2011 11:09 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
...So I'm just at the point where I'm trying to integrate head tracking into the system. Got a bunch of code written to pass tracker data through my system and into GlovePie and I'm ready to test it. I fire up my Vuzix 920VR and start GlovePie and POW! I get the old "Vuzix hardware is no longer supported due to them banning me....yada yada. please help me show those mean old meanies at Vuzix that they should not do bad things to me..." message.

I would be ok with it if he was trying to protect some kind of property rights or something legitimate, but holding everybody hostage because of a pissing match is just stupid. Here I am trying to do some legitimate research for the good of the community, and now I have to waste my time going back and writing my own emulation software because Kenner got his feelings hurt. Thanks Karl. You are a child. :evil:


Fri Dec 09, 2011 12:13 am
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Aug 21, 2009 9:06 pm
Posts: 1644
Reply with quote
I agree, but Vuzix is being just as bad. They should just unban him already.


Fri Dec 09, 2011 12:48 am
Profile
3D Angel Eyes (Moderator)
User avatar

Joined: Sat Apr 12, 2008 8:18 pm
Posts: 11364
Reply with quote
Supposedly the next version of GlovePIE will remove the restriction on Vuzix products. But I agree, the whole thing is stupid.

brantlew: You can try using my software for mouse emulation, should work with 920VR no problem: viewtopic.php?f=120&t=13801

_________________
check my blog - cybereality.com


Sat Dec 10, 2011 3:50 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
cybereality wrote:
Supposedly the next version of GlovePIE will remove the restriction on Vuzix products. But I agree, the whole thing is stupid.

brantlew: You can try using my software for mouse emulation, should work with 920VR no problem: viewtopic.php?f=120&t=13801



Thanks, but I would need the source code. I can't use separate tools for head orientation and position because the calculations sort of depend on each other and need to be synchronized. The emulation is pretty straight forward and I'm about 1/3 through it. I was just annoyed that I had to do now. It's a feature I was going to add anyways - just not this early.


Sat Dec 10, 2011 4:08 pm
Profile
3D Angel Eyes (Moderator)
User avatar

Joined: Sat Apr 12, 2008 8:18 pm
Posts: 11364
Reply with quote
Believe me, I hear that. Because of the GlovePIE/Vuzix war, I had to code support for the Vuzix tracker (which was pretty easy) and then also for the Wiimote (which I wasn't expecting). It was fun, but an unexpected annoyance.

_________________
check my blog - cybereality.com


Sun Dec 11, 2011 1:52 pm
Profile
One Eyed Hopeful

Joined: Mon Dec 22, 2008 4:16 am
Posts: 30
Reply with quote
Have you checked that ?

Scalable Optical Tracking - A Practical Low-Cost Solution for Large Virtual Environments

http://research.edm.uhasselt.be/~smaese ... tions.html


Mon Dec 12, 2011 3:33 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
cbwan wrote:
Have you checked that ?

Scalable Optical Tracking - A Practical Low-Cost Solution for Large Virtual Environments

http://research.edm.uhasselt.be/~smaese ... tions.html



Interesting. The advantages of the system are that the positional accuracy is the same everywhere because the light source is always a constant distance from you and also the simplicity of it (rope lighting is cheap and simple to work with). However it seems a bit of an information overkill. I would think you could do almost as well with 1/10 or maybe 1/100 the number of light sources if they were a bit brighter.

I have considered something similar to this approach - mounting the camera on the head and positioning lights horizontally in the environment instead of the other way around. There are some significant advantages to it in terms of setup and calibration, but there are also some disadvantages with reliability and scalability. It's difficult to judge which is the superior approach without some field testing. I am hoping by the end of January I will have a better feel for it.


Mon Dec 12, 2011 10:15 pm
Profile
One Eyed Hopeful

Joined: Thu Dec 15, 2011 7:17 am
Posts: 3
Reply with quote
brantlew wrote:
cbwan wrote:
Have you checked that ?

Scalable Optical Tracking - A Practical Low-Cost Solution for Large Virtual Environments

http://research.edm.uhasselt.be/~smaese ... tions.html



Interesting. The advantages of the system are that the positional accuracy is the same everywhere because the light source is always a constant distance from you and also the simplicity of it (rope lighting is cheap and simple to work with). However it seems a bit of an information overkill. I would think you could do almost as well with 1/10 or maybe 1/100 the number of light sources if they were a bit brighter.

I have considered something similar to this approach - mounting the camera on the head and positioning lights horizontally in the environment instead of the other way around. There are some significant advantages to it in terms of setup and calibration, but there are also some disadvantages with reliability and scalability. It's difficult to judge which is the superior approach without some field testing. I am hoping by the end of January I will have a better feel for it.


Nice to see you are also interested in wide area tracking. I'm the author of the paper Scalable Optical Tracking. The system could work with less led lights, it just requires the detection of 2 sets of parallel lines. So when working with LEDs, at least 3 lines are needed and the distance between LEDs needs to be at most half the distance between the lines.
A bigger issue is the lack of global positioning. So if you lose line-of sight, move and continue tracking, the system does not detect any movement bigger than the distance between lines.

I'm currently working on a spatial coding scheme to encode the global position. The system loses the indefinitely scalability, but dependent on the code length it could encode entire football fields.

However this system requires some kind of ceiling to mount the LEDs. Putting them on the ground could work, but it becomes a tripping hazard.
I'm interested what tracking solution you're going to chose for your system.

Steven


Thu Dec 15, 2011 11:13 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
@smaesen: Hi. Good to have an academic with experience in this area to bounce ideas with and for advice.

smaesen wrote:
I'm interested what tracking solution you're going to chose for your system.


I have an open mind right now as to how I am going to implement. Since this is targeted for a consumer-grade gaming system I don't need sub-centimeter positioning, so I have a lot of freedom with regards to technology. I will almost certainly be using a combination of optical and inertial systems. The inertial system will be for head-tracking and potentially for fine local motion. The optical system will track low resolution absolute position and is necessary to correct errors in the inertial system, to enforce real-world boundaries, and to implement redirected player movement.

Right now I don't know if the best place for the camera is mounted at a high vantage point viewing the player or mounted on the players head like your system. There are tradeoffs in accuracy, scalability, and convenience that I need to investigate. Mounting a camera on a pole requires bulky equipment and a lot more calibration - compounded by the number of additional cameras. However once calibrated the system is very simple, reliable, and scales easier.

Mounting on the head seems like it can simplify the setup. I was thinking of using some type of optical markers placed around the playing area (lights in a known triangular configuration) that when visible could be used to deduce the global position. (is that what you refer to as "spatial coding scheme"?) That would simplify the setup and calibration a lot but would not be as reliable since there would be "blind-spots". Also to enlarge the area would require enlarging the markers and/or increasing the number of markers so there are scalability problems.

Anyway these are just my speculations and ramblings for now. Like I said, I'm hoping that some field testing will give me some guidance. Maybe you have some opinions on this since you have some practical experience with it?


Thu Dec 15, 2011 2:54 pm
Profile
One Eyed Hopeful

Joined: Thu Dec 15, 2011 7:17 am
Posts: 3
Reply with quote
It's a good idea to combine the optical and inertial tracking systems to get both local and global tracking. As you will notice, your inertial system will drift over time. To correct this, you can use the less accurate optical tracking to merge those 2 streams together. But care must be taken not to introduce too much jitter from the inaccuracies of the global tracker.
To correct for inertial drift of the orientation, u need to use a sensor with a compass to correct. Else it will be difficult to redirect someone if you're not certain which direction he is facing.

Maybe a good question to ask yourself is: do you want a system to be used outside or inside. Outside you have more space, but cameras have it more difficult due to the sun and other visual distractions. It also restricts your control of the environment (no ceiling :-), bumby ground, obstacles, sun, weather, ...). But inside you lack the space to move freely, unless you live in a castle ;-).

Outside cameras looking in is a good way to get a global position. But as you said, it has problems with scalability. Accuracy drops linear the farther you move from the camera. Resolution from the tracking space can be easily calculated with the resolution of the camera, FOV of the lens and distance to the camera (Pythagoras). Also what you need to take into account is the processing for the global position needs to be done as close as possible to the camera (to decrease latency) and then sent wirelesly to the gamer.

Alternatively, camera mounted on the gamer gives you a good global orientation as well as a global position (if the distance to the global markers is kept limited as in my paper). But it needs to see enough markers in each frame to get an accurate result. In an open field, this could be a problem. You're indeed correct that you need to scale the markers with your environment as a single triangle can not be seen from every position. Therefor you need to encode the global position into these markers so that when you view a subset of them, you still can tell where you are (so ID each marker in some way, like spatially or by blinking in a pattern). That's what I meant with the coding scheme.

I'm also rambling about tracking because there is just so much to tell :-) But field testing it will certainly give you a better insight. Try to define your problem or setup a bit more so you can experiment more directly.

PS: also cbwan can give you a lot of advice on good cheap hardware and DIY VR setups. You probably know his webblog 'VR Geeks'.


Thu Dec 15, 2011 4:37 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
smaesen wrote:
Maybe a good question to ask yourself is: do you want a system to be used outside or inside.


Thanks for the advice. Its good to have an expert's input.

Right now I am targeting an outside environment, because I just don't think that a natural motion system can be accomplished well within the constraints of everyday structures. A gymnasium could be an exception but for most people, outside areas are more readily available. In my mind, the key to the system is redirected motion. If the player can operate for extended lengths of time without crossing a boundary then the system is useful - otherwise its just a tech demo. Learning the limits of this technique is my main area of interest. Unfortunately I have to build up a lot of support technology before I can even start that part of the project. :)


Thu Dec 15, 2011 5:18 pm
Profile
Certif-Eyable!
User avatar

Joined: Sat Dec 22, 2007 3:38 am
Posts: 990
Reply with quote
For the Global position, if using the rope lights, could you use two rows of 4 lights in a corner of the square to determine the position.
With 4 lights, you can mark the position within a 16*16 grid.
The reason I say a corner is so that you can tell the direction you are facing too.

The other option is to blank out some of the lights in each sector of the lights:

o_o_o_o_ooooo|o_o_o_oooooo|o_o_ooo_ooooo etc


Thu Dec 15, 2011 5:32 pm
Profile
One Eyed Hopeful

Joined: Thu Dec 15, 2011 7:17 am
Posts: 3
Reply with quote
android78 wrote:
For the Global position, if using the rope lights, could you use two rows of 4 lights in a corner of the square to determine the position.
With 4 lights, you can mark the position within a 16*16 grid.
The reason I say a corner is so that you can tell the direction you are facing too.

You could indeed add coding lights to every square in the grid. But it adds more lights to place, to process and buy. Also with 4 lights you can only encode 16 places, so a 4 by 4 grid

android78 wrote:
The other option is to blank out some of the lights in each sector of the lights:

o_o_o_o_ooooo|o_o_o_oooooo|o_o_ooo_ooooo etc

That was indeed an other coding system I considered. In the end I reduced the number of LEDs even more by stepping of the idea of a grid and just used coded (on/off) parallel lines in 1 direction. Also turns out that using only 1 color speeds up detection by a factor of 10. I'm still working out the details and hoping to test my new tracking system in the near future.


Thu Dec 15, 2011 6:23 pm
Profile
Certif-Eyable!
User avatar

Joined: Sat Dec 22, 2007 3:38 am
Posts: 990
Reply with quote
smaesen wrote:
...
You could indeed add coding lights to every square in the grid. But it adds more lights to place, to process and buy. Also with 4 lights you can only encode 16 places, so a 4 by 4 grid
...

Yes, but two rows of four lights would be 16*16. ;-)


Thu Dec 15, 2011 6:49 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Jul 08, 2011 11:47 pm
Posts: 1497
Reply with quote
I'm sruprised someone hasn't just hacked this exclusion out of GlovePIE. Unless he's put some effort into protecting it, you can probably just NOOP out the instructions that do the check...


Thu Dec 15, 2011 7:21 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
Project Update:

I'm about a month behind where I thought I'd be by now because of the holidays and having to write a mouse emulator, but I'm back on track now and hit a major milestone tonight. I finally was able to scale up the system and physically walk around SkyRim in my backyard! :D

I need to qualify that sentence because it sounds a lot better than it actually was. I was confined to a very small 15' x 15' area (maybe double that area in-game) and there were a ton of calibration problems so I had to strictly regulate my walking speed and keep my head still while walking, etc.., but even so - the basic mechanics did actually work properly. I could look around, turn, walk forward and backward, strafe, walk forward with my head turned sideways. Everything was translated into the correct game motions - and for a few brief moments it was sort of surreal.

However the numerous bugs and calibration problems kept it from being very immersive. Here are some major problems that I identified.

1. Numerous calibration and calculation issues. Most of these were already known, but I have been plowing ahead without fixing them because I wanted to get to this milestone quickly. Now I can go back and clean up.

2. I rigged a quick 6-LED light source together and didn't bother to improve the emission FOV so the camera would lose sight of me quite often. One thing that I learned from the experience is how problematic head motion is going to be for the LED camera because it tends to move the light into an edge on orientation. Also looking way up or down occludes the light with your head. I definitely need to create a very diffuse light source.

3. A fundamental problem is the total "disconnect" between your real life walking speed and the character speed. All the subtle speed variations that you tend to make are completely lost and translated into a binary "stop" or "go" speed. That is a major inhibitor to the feeling of immersiveness and is potentially a deal-breaker for the whole idea - at least for generic gaming.


Tue Jan 03, 2012 12:47 am
Profile
Sharp Eyed Eagle!

Joined: Mon Aug 09, 2010 6:08 pm
Posts: 381
Location: Houston
Reply with quote
brantlew wrote:
I could look around, turn, walk forward and backward, strafe, walk forward with my head turned sideways. Everything was translated into the correct game motions - and for a few brief moments it was sort of surreal.

You took a brief step forward in time. That feeling of being in the world is what people of the future will constantly experience.


Tue Jan 03, 2012 1:22 am
Profile
One Eyed Hopeful
User avatar

Joined: Thu Dec 29, 2011 1:15 am
Posts: 44
Reply with quote
Branntlew... thats pretty awesome man. KUDOS to you!

Physically walking being transfered into a digital world is pretty sweet stuff. I think Omni Directional Treadmills are the way to go for this type of tech. Once they become more compact ofcoarse.



_________________
PSN/XBL/NNID/STEAM: Moggle69


Tue Jan 03, 2012 1:58 am
Profile
Certif-Eyed!

Joined: Tue Jan 19, 2010 6:38 pm
Posts: 529
Reply with quote
WiredEarp wrote:
I'm sruprised someone hasn't just hacked this exclusion out of GlovePIE. Unless he's put some effort into protecting it, you can probably just NOOP out the instructions that do the check...


Sure, probably wouldn't be that hard, but if someone did that, and released it publicly, we might just end up with all 3d devices banned.

brantlew wrote:
Everything was translated into the correct game motions - and for a few brief moments it was sort of surreal.



Heh, Strange, isn't it :)

_________________
"If you have a diabolical mind, the first thing that probably came to mind is that it will make an excellent trap: how do you get off a functional omni-directional treadmill?"


Tue Jan 03, 2012 6:00 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
Thanks for the encouragement guys. It helps keep me motivated on a long project like this.

@Moggle69: Omni-devices are cool but I don't think they are very realistic for the average guy. It's bad enough we have to drop over $500 on an HMD, but adding in a piece of equipment that would easily cost over $5K is too much to ask of all but the most die-hard guys. A natural motion tracking system could potentially cost less than $200 and provide a superior experience than any omni-device! So from a pure "value" perspective I think some type of tracked natural motion system (whether Kinect or something like mine) is a clear winner. If you want to experience VR movement anytime soon (outside of CES demos) then this may be the only option.



I've been thinking about problem #3 a bit and I think maybe using the 360 controller interface instead of WASD keys would help a lot. The 360 controller provides continuously variable character speed and direction - true? Unfortunately since GlovePie 0.45 STLL doesn't support Vuzix I'm going to have to implement yet another device emulator if I want to test that. Aaargh! (or get a different HMD ;) )


Tue Jan 03, 2012 1:40 pm
Profile
One Eyed Hopeful
User avatar

Joined: Thu Dec 29, 2011 1:15 am
Posts: 44
Reply with quote
brantlew wrote:
Thanks for the encouragement guys. It helps keep me motivated on a long project like this.

@Moggle69: Omni-devices are cool but I don't think they are very realistic for the average guy. It's bad enough we have to drop over $500 on an HMD, but adding in a piece of equipment that would easily cost over $5K is too much to ask of all but the most die-hard guys. A natural motion tracking system could potentially cost less than $200 and provide a superior experience than any omni-device! So from a pure "value" perspective I think some type of tracked natural motion system (whether Kinect or something like mine) is a clear winner. If you want to experience VR movement anytime soon (outside of CES demos) then this may be the only option.


I hear ya.... my only question is, how do we fix the barrier issue?

Lets say you have an empty 200sqft room. You start in the center of the room, put your HMD on and start playing. What is going to prevent us from walking into the wall?

_________________
PSN/XBL/NNID/STEAM: Moggle69


Tue Jan 03, 2012 7:08 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Jul 08, 2011 11:47 pm
Posts: 1497
Reply with quote
Quote:
Sure, probably wouldn't be that hard, but if someone did that, and released it publicly, we might just end up with all 3d devices banned.


I thought it was Vuzix that had banned Carl Kenner, but I see in the documentation.rtf for GlovePIE it seems to have been 3Dmtbs.com that banned him?

Quote:
Stereoscopic 3D

Stereoscopic 3D support has been removed in protest against MTBS3D.com (the Stereoscopic gaming organisation) banning me.

So Cam.Stereo, Cam.ScreenDepth, and Cam.EyeSeparation no longer work.


Is this old or something? Or has he been banned from heaps of places for some reason?


Tue Jan 03, 2012 7:19 pm
Profile
Sharp Eyed Eagle!

Joined: Mon Aug 09, 2010 6:08 pm
Posts: 381
Location: Houston
Reply with quote
Are you quoting Carl Kenner? Is he banned from MTBS?


Tue Jan 03, 2012 7:36 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Aug 21, 2009 9:06 pm
Posts: 1644
Reply with quote
Moggle69 wrote:
I hear ya.... my only question is, how do we fix the barrier issue?

Lets say you have an empty 200sqft room. You start in the center of the room, put your HMD on and start playing. What is going to prevent us from walking into the wall?


I suggest you take a look at this thread: viewtopic.php?f=120&t=13784

As for avoiding hitting the walls, the answer is re-directed walking. You will find some stuff in that thread, I believe.


Tue Jan 03, 2012 8:01 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2217
Location: Menlo Park, CA
Reply with quote
There's also a good bit of discussion of redirected walking throughout this thread.


Tue Jan 03, 2012 8:56 pm
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 184 posts ]  Go to page 1, 2, 3, 4, 5  Next

Who is online

Users browsing this forum: No registered users and 4 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
Powered by phpBB® Forum Software © phpBB Group
Designed by STSoftware.