Home built Omni Directional treadmill

[Applemung]

Hey guys,

I'll start off with a bit of a background.
I have always been intrigued by virtual reality but the technology and the gaming industry were always worlds apart.
There was hardly any options and with what there was available it didn't seem to capture my imagination.
Last year I viewed a video about the ultimate BF3 simulator (http://www.youtube.com/watch?v=eg8Bh5iI2WY)
and was amazed by the immersion and set about to find a way to bring it back home. Alas this was not to be so.
After much pondering I came to the conclusion that the picture being beamed onto the wall was going to be the most
difficult part to re produce. It seemed to me that it was not worth the effort because of how limited you were in looking
around the environment and I felt any immersion would be lost once the initial novelty of the system wore off.
I looked at many head mounted displays etc but nothing seemed to tick all the right boxes so I gave up.

To me, a newcomer to the VR scene, a whole world has been opened up with the oculus rift. Even in it's infancy
it has generated soo much buzz for the non VR community that I simply had to investigate it further. I hadn't heard of the
oculus kickstarter campaign and was a bit annoyed I missed the initial backing orders. I know the resolution is poor and the
screen door effect does not bother me in the least, your brain can teach you to ignore such imperfections.

So, now to the project. The last of the devices for fully immersible virtual environments is a way to move around your environment
without actually moving (or as little movement as possible). I'd searched around the net for an easy, affordable solution and found a
stack of them. But yet again none of them was to me, worth the effort. I've seen the virtuix and similar but they still lack "natural movement"
that I have been looking for.

With no other options and no foreseeable consumer omni treadmill in sight I decided that I will build one in my shed. The design is going to follow
the omni directional treadmill created by MSE as seen in the BF3 simulator. I picked that one and not cyberwalk because of complexity and cost.
Whith an estimated cost of about $5000 AUD it would be significantly cheaper than the cyberwalk. That is mostly just materials cost. The bearings alone are going to total
$2500 . That does not bother me though as this omni is going to take a while to build so the costs are going to be projected over quite a period of time.
After all who else in the world has one of these in their backyard? I'd love to see one if anyone has a link!

So at the moment I am in the design/plan and test phase. I have drawn a quick sketch sized to fit in my 3.5 meter wide shed. Total outside diameter is 3 meters.
I used a free version of solid edge 2d to draw the sketch. The sketch shows some simple but not definitive dimensions, 16 treadmill roller segments with one segment
showing a few dimensions of roller sizing. The center piece will be a solid platform for the user to stand on with the scetch showing two of the support beams drawn in.
The sketch is no where near complete but it at least gives me an idea and shows any possible problems with design.

Please feel free to critique and advise on any aspect during the design phase as many eyes are better than two!

Because I hate paperwork I decided to get my hands dirty and put together roller mk1!
The rollers are going to comprise of a center support shaft that the rollers will spin around. The bearings will sit on this shaft held in place by collets secured to the shaft on the inner side
of the bearings (the two nuts on the threaded rod in pic1) and a thin spacer on the outside of the bearing (the nut on the threaded rod pic2).



The hardest part with the roller is finding a bearing/roller combo that fits. With standard bearing sizes and standard tube sizes I have very limited options. I have chosen a 32mm tube 1.6mm thick
and mated it with a 28mm bearing. This leaves a gap of 0.4mm on the outside of the bearing. I plan on using a thin plastic sleeve to make up for the space left over held in with an epoxy resin.
This quick test rolelr has proven that material available locally to me can produce a very nice and tight solution. In the pic I have used threaded rod because it was just easy to use and I had it
laying around.

My next goal is to create one section of this treadmill and actually have it powered.

I'm loving this project already!

[brantlew]

Cool project! Please post plenty of progress photos.

[blazespinnaker]

Hmmm..

I watched the video a couple of times. I noticed while he was walking sideways across the ODT, the ODT would be sliding him towards the center (the force vector is orthogonal to the direction he's walking). I think I'd fall pretty fast on something like that, especially if my eyes were covered by the rift. The thing about the BFE video is that your eyes aren't covered (you can see yourself relative to the ODT).

One of the problems with ODT + rift covering your eyes is the need for a harness, especially if you want to be able to jump and crouch, etc. The more active you are, the more likely you're going to lose your balance. But being active is what makes the ODT so much fun.

In some ways, the rift is worse than being blindfolded. At least with a blindfold you don't have these contrary cues coming at you telling you something different than what your inner ear is telling you.


Here's an interesting survey of ODTs out there, if you're looking for some ideas:

whole.pdf

[Applemung]

Hey, thanks for the input! That research paper was quite interesting but a tiny bit outdated, very helpful nonetheless.
At first it seemed to me like the motion change was going to be a deal breaker but after a while of pondering
I came to the conclusion that a user of the omni would get used to the motion and walking style when on an omni. You would be able to
train yourself to adopt a walking pattern simply by walking on the platform without any HMD and get used to the inertia changes when turning.
The speeds we are talking about are from a cautious walk and brisk walk. The size of the treadmill would not permit full blown running and I would
need to upgrade the size to a whopping 5 - 6 meter radius to be able to speed up the motors fluidly enough. Crouching would be extremely easy, as
you generally do not go into a crouched position immediately during walking. You would slow down momentarily and then begin to descend at which point
the rollers would have slowed down enough not to put you off balance. Of course this is all just speculation but I'm not going to be 100% sure until it is tried and tested.

One thing I had in mind was that when the system sensed that you changed direction it would slow the rollers down until you were facing 0 degrees again
at which point it would speed up the rollers. I am hoping that I can get sensors to track fast enough to allow quick adaptations to roller speed.
I think in this day and age it should be very achievable with the amount of sensing options out there. Even if I had to couple many different types in the one system.
Inertia sensors was one thing that popped into my mind.

I think that this treadmill should be excellent when coupled with the rift but I think I should de a bit more reading before fully committing to the system build.

[blazespinnaker]

It's true, you do train yourself, however certain movements (crouching and jumping) which are a lot of fun on an ODT are very hard to train to a point you can do them safely without a harness.

[Applemung]

Yeah I think jumping is going to be the biggest issue and I don't really think that there will be any way at all that you could safely jump without a harness.
I think crouching you could also get used to. One thing I was thinking of was a harness system that follows a roller track half the diameter of the omni treadmill suspended from the ceiling. This would give you unimpeded movement around the treadmill but ensures that if you fall that you won't hit the ground or at least hitting it very softly by use of short bungee cords. I more so fear damage to components rather than damage done to myself.

I was thinking of designing a game that suits the oculus rather than playing games that were more so suited to movement and control type of keyboard and mouse.
This guy is a freaking genius and I will be following him very closely. http://idav.ucdavis.edu/~okreylos/ResDev/

One of his most interesting videos http://www.youtube.com/watch?v=R0-dsbeasgA

[geekmaster]

... This guy is a freaking genius and I will be following him very closely. http://idav.ucdavis.edu/~okreylos/ResDev/

One of his most interesting videos http://www.youtube.com/watch?v=R0-dsbeasgA

His username here is doc-ok, and here is one of his posts:
http://www.mtbs3d.com/phpBB/viewtopic.p ... 05#p107905.

Here are his blogs about the Oculus Rift, Razer Hydra, and other things:
http://doc-ok.org/?p=548
http://doc-ok.org/?p=493

[Applemung]

Oculus Rift has been ordered! Now to play with Unity 3d and make an fps suited for the oculus in parallel to making the ODT! happy in the pants.

[geekmaster]

Oculus Rift has been ordered! Now to play with Unity 3d and make an fps suited for the oculus in parallel to making the ODT! happy in the pants.

That is called PTSB.

[Applemung]

Progress report.

I have made up 4 rollers as a bit of a proof of design and they seem to work brilliantly do far. I ended up using heat shrink over the bearing to make up for the 0.4mm gap between the bearing and the roller. Because of the tight fit I no longer need the collets to hold the bearings in place as the epoxy will do just fine. That should save quite a few kilos in weight on the overall system. I can stand on the rollers and slide along them quite freely and it feels sturdy. I have threaded the end blocks so that the center shaft stays put and does not freely slide. The end blocks will be re made for the final product and have an extra cut added at an angle at the ends so that the rollers from each bank can sit snugly up against each other. I think I am now at the stage where I am happy to purchase enough materials to make a full bank of rollers. I'll be looking out for a treadmill to turn into the roller tread.




In addition to my treadmill build I am learning to program using unity and the integrated scripting program that it has. The only experience I have had with programming was back in school where we used machine language to get a Z80 chip to do a washing machine cycle. Oh that and about 40 lines of code in Arduino. So it is quite a steep learning curve but I am more than confident in nutting it all out.
I have so far created an simple fps where I run around shooting projectiles in a basic environment with a gun in the shape of a stick. In the same breath I am learning 3D modelling with Blender and am currently about 20% done on my first fps gun model. Unfortunately progress is slow as I only have about 4 hours solid in a day to do everything. I'm kind of holding off going further with Unity until I get a Razer Hydra and my Oculus Rift so that I can figure out the camera controller and gun control. I am at a stage where I feel comfortable in creating the environment but not too sure about game mechanics. Plenty of tutorial videos to watch. It's annoying going back and forth from Unity and Blender as the controls are polar opposites.

Back to the grind.

[Applemung]

Small update,

I might just be the person that has waited the least amount of time for the rift. Ordered july 6th and arrived 23rd.
And to jelly further, I am in Australia. Weird, but lucky.

My first impressions. I am sold. Take my money. Big plans, angry missus. profit??
Ok more seriously. As a newbie to VR I could use this HMD as an end product despite it's shortcomings.
Firstly the low resolution:
I grew up with a really crappy first computer, that was already old back in counterstrike 1.3 days.
If any of you guys played that you would know that the graphics even on high were not state of the art, so divide that by 10 and that's how I played
for the first year of my PC gaming life. Prior to that I was an avid Super Nintendo player. In short I am used to low resolutions so that does not affect me
whilst using the rift. I have found that forcing AA and playing around with the resolution has sharpened some aspects of rift demos.
The screen door effect:
What screen door effect? Seriously I can not fathom why people are finding ways to diffuse the screen to minimise the already minimal effect. I can only see the screen door effect when I try to focus on a single pixel and only then it becomes apparent. Maybe I'm still caught up with the whole immersion effect that this device gives.
I know that science has already proven that our brains wash out persistent image effects, maybe my brain does this a little quicker or at least I don't focus on the
imperfections so they don't seem apparent to me. The first imperfection I saw was the image blur when turning around quickly but even that seemed to disappear after a minute.
The sickness:
The only time I felt motion sickness was was when I was doing things that I know should not have been done ie controlling head movement with a joy pad along with the rift tracking. Any other time where I only control forwards and backwards feels comfortable. The bummer comes a day after use. Throbbing headache this morning, but not immediately after waking. I was fine when I first got up but about an hour later I went to pick something up off the ground and then it felt like all the blood rushed into my brain and BAM! Throbbing headache for half a day. Not good when my job requires precision and fine attention to detail. Good thing I'm inundated with paperwork at the moment cause id feel like I'd make a very expensive mistake otherwise. I'll try using some other lenses and see how it goes. Maybe its from my lack of never using any 3d product before, maybe only with time it will go away so for now I'll take it 15 mins a day and see if the headaches get better or worse.

So the to do list:
Get Razer Hydra
Get MS Kinect x2 for leg movement detection
Ultrasonic positional detection using echo location
Continue with ODT construction
Design of new FPS VR gun
Continue to learn the ways of Unity and programming in general
Nice to do: Make Oculus Wireless (not a high priority at the moment)

About the gun: One of the immersion busting things I have noticed in most fps based VR demos is guns have no recoil or more so there is no effect on your aiming precision when firing a gun. While there might be a 3d animation of recoil your cross-hair remains the same and you can either just continue to fire or wait for the cross-hair to settle in case of time based accuracy (less gun accuracy if button is being mashed). So I came up with a fully workable and safe idea to replicate what it would feel like to actually fire a gun and have it actually throw off your aim rather than software based. The gun would be a rifle style. By using solenoids and weights I can simulate to an extent, recoil. Imagine cutting a gun stock in half and then placing a solenoid in between the two pieces. As you fire the gun the force of the solenoid would extend the rear half of the stock effectively applying some recoil force onto your shoulder. The rear section would also have weights placed inside so that in case you wanted to one hand the gun away from your shoulder the solenoid action would still throw your aim out.

[android78]

So, looking at that to do list, you're almost there?
I did have to chuckle at 'I grew up with a really crappy first computer, that was already old back in counterstrike 1.3 days.'. Guess you don't remember being blown away by the ability to run around a 3D castle when you first played Wolfenstein. Imagine what it was like when doom came along with multiple floor and roof levels, not to mention walls at all angles... WOW! I remember making designs for suspending a CRT from the roof and use a ball mouse to track turning so I could play doom first person. Didn't come to fruition, however the mouse was never the same after pulling it apart, wrapping string round one of the tracking rotors and testing head tracking with it.

[Zoide]

It's been a while since the last post. Any news?

[Applemung]

Should be some updates this week or mid next week, I've changed the way that the rollers are secured onto the sections and am getting some materials to try it out.
Been extremely busy at home and at work so it has not had that much work done. I've Put together part of a frame for one section, made a crude base plate, bought
a second hand treadmill and partially disassembled it, bought an extra 40 bearings to get the concept going and all I need to get now to finish off the section is some 25x30x3mm aluminium angle.
Hopefully by the end of the week or mid next week I should have one section in some sort of functional state. I've only got two days, with about an hour split between them to do it.

Cheers!

[Applemung]

As promised, I have made some progress with the build. I have so far put together a section and roller support bracket for about half of the rollers on one section. This should at least give me an idea on how the system will work and what issues I might come across. In the pictures you can see how the shafts are supported on the brackets. This is quite a rough placement though as I was just recycling things I already had. The rollers will have short spacers either side placed over the shaft to stop any side to side movement, Solid rod will be used in place of threaded rod held on both sides with either a split pin or circlips. I've stood and bounced over the rollers and it feels quite strong. The way that this has been constructed should enable me to put together each roller segments separately and then bolt them on individually over the frame. I plan the have the roller shafts sitting on a rubber grommet in the holes so that when you step on a roller it will compress onto the tradmill and gain better friction, by doing that it should also negate any miss-match on roller height differences.




Under view


Over the next week or so I will finish off the segment and butcher the treadmill to power the rollers in some crude fashion that should enable some sort of vague movement.
Now that I'm getting further in the building process I'm getting more and more motivation to get the project complete.

[colocolo]

that looks good!
any estimations about material costs? looks a bit costy.....
have you already tried to stand on those first mounted rolls to see if they roll?

[Applemung]

Yeah, even without the spacers stopping the side to side slop they roll very freely.

So far it has cost me just over $300, I have enough materials to half populate this section, about another $100 to finish it off.

Costs so far include:

40x40x1.6mm ally square tube
32x1.6 ally round tube
various ally angle
nuts/bolts
second hand treadmill
28mm bearings 6000RS
12mm Threaded rod
12mm Solid ally rod

[JohnyX]

Yeah, even without the spacers stopping the side to side slop they roll very freely.

So far it has cost me just over $300, I have enough materials to half populate this section, about another $100 to finish it off.

Costs so far include:

40x40x1.6mm ally square tube
32x1.6 ally round tube
various ally angle
nuts/bolts
second hand treadmill
28mm bearings 6000RS
12mm Threaded rod
12mm Solid ally rod

$400+/- per section ?

[JohnyX]

Have you looked in to recycling used conveyor lines ?

[Applemung]

I'm looking to get as much recycled parts as I can to cheapen the cost but down here in Adelaide Australia there isn't such a huge market or advertising for such products.
There might be a few bits and bobs lying around that could be cobbled together but generally it is cheaper to just get the exact part you need. Yep, that's about on the money for $400 +/- per section with the biggest single item cost being the bearings. In reality, most people can actually afford something like this (if it's home built) if you look at the time scale that it will take for me to finish with costs being projected over time. You could say that this is costing me about $80 per week. That's about the same price that I am paying of my solar power system repayments. Get a few mates to pitch in as well and you could have it even cheaper, just comes the issue of who keeps it and how to share . So far I'm riding the costs alone but I don't mind.

This is definitely not the omni treadmill for every home but it is certainly quite a rare omni to have.
Personally I am really excited in getting this finished but it's going to take me quite a while to fully complete.

[Applemung]

Yes I'm still alive and the project is moving along with a little bit more pace now.
My workload increased by about 100% as we near the end of the first phase of our project (Design/Test/Evaluate/Break/Fix).
Every time we complete one thing, 10 more things appear out of the blue. Things are calming sown a bit though so I found some spare time.

I've also been busy setting up a computer at work to play with the rift and other toys, in doing so I ended up modding the rift
to minimize the screen door effect (I like to mainly use the B/C cups so it is more apparent). I placed a privacy screen filter about 1mm
from the screen using sticky plastic strips and scotch tape. While the screen door is still there is is extremely faint and the image quality is
much better. I thought that there would be a trade off by placing the filter in but so far it has only been positive. It literally feels like I have put in a higher pixel
density screen in. A good example of the quality is best seen in the Tuscany demo when you are looking at the leaves in the trees. With the original screen
you could clearly see the screen door effect quite badly when looking at the leaves and moving your head about. With the modded screen it just looks like it is
a low resolution model. Every game just looks so much better.



Privacy screen on


In addition I have been working on the Omni treadmil. I have re designed the internals of the rollers for cheaper/easier construction.
I have bought 12mm aluminium rod, 4mm split pins to stop the bearings from sliding side to side and also as the outer stops in the cradle (Not shown). I have also bought some
16mm plastic tubing (12mm ID) as the spacer between the bearing and cradle (Not shown). I'm calling the aluminium angle the cradle for the rollers.





I will be picking up the pace for the next few weeks as I have been gathering more and more materials for the project. I am also nearing a design for the belt drive to power the first section.
I hope that by the end of the month I will have one powered section near completion or at least in a state that will tell me that the project is going to work or not.

[mayaman]

Has anyone used my 3M privacy filter trick on the DK2?

[emsman]

I know of one other ODT out there that has a good design.
Infinadeck.com