Envisioning a DIY CyberCarpet for Omnidirectional Motion

Post by **pierreye** » Thu Oct 20, 2011 8:12 pm

We could use undercarriage spray that will coat thin layer of rubber on the steel ball to increase resistance.

Post by **brantlew** » Thu Oct 20, 2011 8:25 pm

@Cyber - You wouldn't need to attach wiimote to shoes, just tiny IR LEDs. The WiiMote is the camera that tracks them. Still - there is some work that would need to be done to translate those moving lights into accurate 3D coordinates. I say just use the Kinect and grab those coordinates for free, plus you get all the other body position information along with it like hip placement, torso orientation, head height, arm placement which could be useful for other things.

I'm curious if the Kinect lag is due to the processing power of the on-board Kinect computer or the sensors? The OpenNI demo was interesting because it really seemed to be accurate and real-time. So maybe the solution to lag is bypassing the Kinect CPU and doing those computations on the main computer?

All of a sudden I've got an itch to buy a Kinect now.

Post by **brantlew** » Thu Oct 20, 2011 8:28 pm

pierreye wrote:We could use undercarriage spray that will coat thin layer of rubber on the steel ball to increase resistance.

Good idea. I wonder if the rubber would just rub off though. How about using plastic or even solid rubber balls?

Post by **cybereality** » Thu Oct 20, 2011 8:39 pm

@brantlew: I wasn't suggesting the use of the IR camera on the Wiimote. It also includes a 3 axis accelerometer, among other things. It tracks relative motion. You actually don't need the absolute position. All you need is that vector between the two feet, which the accelerometer and some simple math would provide.

Post by **brantlew** » Thu Oct 20, 2011 8:50 pm

@Cyber: ok, I see where you are going with that...

Post by **brantlew** » Thu Oct 20, 2011 9:15 pm

What about using steel balls but putting rubber matting on the inside of the dish? The same interlocking mats they use in weight gyms. As a bonus - it would cushion your fall

Post by **mAchiNE** » Thu Oct 20, 2011 9:26 pm

If mounting the ballbearings to the floor and walking on them is ok then mounting them to your shoes and walking on them should be exactly the same, the amount of friction between you and a stable surface is the same in both cases.... what it all comes down to is how much friction those Ball Transfer Units actually require to move them.
Coating the balls in anything probably wont work because the casing would scrape it off again and if any bits got stuck inside the bearings then you'd have bearings with uneven friction which would cause problems of its own, I think you could spray the rubber onto the curved surface you are walking on to improve grip of the ballbearings to the walking surface which should remove any "slip" where the bearings are just sliding on the surface instead of rolling on it.

EDIT: @Brantlew I think a Rubber mat could work as long as its not too soft

Post by **pierreye** » Thu Oct 20, 2011 9:55 pm

I have some hard rubber mat that use for outdoor playground. I could test it out once I source out the ball transfer units from local supplier. I estimate my cost would be around USD 200 for the whole project.

Post by **PalmerTech** » Thu Oct 20, 2011 10:10 pm

The problem with roller shoes is more complex than it seems. It is definitely not the same as having the balls on the floor!

For one, you would be elevated pretty high above the ground, even with smaller transfer units. Ever try running in platform shoes? Ask your girlfriend, they will explain.

In addition, normal shoes flex quite a bit when you walk, and even more when you run. Having the shoe be a hard, solid platform with balls on it would be pretty hard to walk on/keep your balance on!

I suppose it could work, maybe I will try it out. That bearing supplier I found is the cheapest I can find, and it turns out that they are based less than an hour drive from me, so I could even do local pickup. I don't expect it to work as well as an entire floor, but it might work well enough.

Post by **Synexious** » Thu Oct 20, 2011 11:31 pm

The shoe idea is the only option that is affordable. Tens of bearings are affordable, thousands are not.

Post by **mAchiNE** » Fri Oct 21, 2011 2:28 am

@palmer
have you ever worn military issue boots? they are not all that flexable, its not like soldiers are running around in nike running shoes

but i get your point

maybe we can use these bearings:
http://www.vxb.com/Merchant2/merchant.m ... sfer-Units" onclick="window.open(this.href);return false;
or these
http://www.vxb.com/Merchant2/merchant.m ... sfer-Units" onclick="window.open(this.href);return false;
and counter sink them into some boots with fairly thick soles to keep the height to a minimum sure they will be a bit stiffer than normal shoes but I think its worth a try, the cost saving will be massive if we can use "cyber shoes" instead of a "cyber carpet"

Post by **Synexious** » Fri Oct 21, 2011 4:00 am

The shoes wouldn't have to be flat and firm. They could still bend. The bearings could be attached individually so the shoe would flex. The height would be an inch or two extra, not much. Less than rollerskates.

Post by **brantlew** » Fri Oct 21, 2011 5:32 am

I like mAchiNE's idea of using lower profile bearings and sinking them into thick soled shoes to minimize the height impact. The second unit looks more realistic to me since the first one is so wide that you would not be able to fit many together. These are all really large bearings (1" or so). At that size we could only get a few of them on there and they would really be more like rollers skates. Have you guys run across any smaller size bearings - maybe in the 1/2" range?

Post by **WiredEarp** » Fri Oct 21, 2011 6:01 am

This is a rather wasted thought, but how about a smooth/teflon type surface, under which sits 2 electromagnets on 2 2 DOF frames (like cranes).
The use could wear shoes with metal plates in them, and when he walked forward, the system would 'slide' him backwards by applying force to the foot that is stationary.
You could call it the MoonWalk

Post by **brantlew** » Fri Oct 21, 2011 2:58 pm

I know this is not what we're all looking for, but what about something MUCH more simple and practical. What if you just got a cheap motorless treadmill and a Kinect.

http://www.sportsmansguide.com/net/cb/S ... medium=cse

You would be limited to just forward and backward motion, but it would be very natural and unimpeded. No weird shoes or balls to slip on. No complicated construction. You might attach a safety cable in case of a fall but it wouldn't need to be load bearing. To track speed and direction, you could use some type of mechanical odometer or the Kinect API. The only thing left would be turning. You could use a simple gesture like torso twisting to accomplish that - ie. turn your chest left and you turn left, face forward and you stop turning. That could be tracked either via Kinect or IR sensors.

I know it's not the perfect solution but it accomplishes some of the core goals - realistic locomotion, inexpensive ($350), and trivial to implement. Would the simplicity and price of this system outweigh it's limitations. Personally I believe so.

Update: I'm not sure how much force you actually would need to start the belt rotating. It might not be possible to just start walking and have it turn. Near frictionless rollers would be necessary. Maybe you could mod an existing treadmill or maybe you could build something easily.

Post by **cybereality** » Fri Oct 21, 2011 6:42 pm

@WiredEarp: I had thought of a similar idea, but maybe cheaper and more practical. What if you were in a circular pod (like a metal hula hoop size) and had your feet tied to resistance bands. So you could pick up your feet, but only move forward and back slightly (then it would pull you back into position). The metal hula hoop could spin within the pod, to allow you to freely turn. But the idea would be that you just couldn't leave the circle. You would pick up your foot and "think" you moved it forward, but the resistance bands would absorb all the force and return your foot the the floor in the middle. Seems like it might be dangerous, but I think somehow it could work.

@brantlew: Honestly, that idea (which has been discussed before) is the most obvious, practical, and affordable one out there. The main problem is that its not omni-directional, so you are limited to only moving forward (I don't think you can go backwards on those things). I have tried one once, you can walk at different speeds, you don't have to run. It would still be better than using mouse and keyboard, but I am not sure it would be any better than using a wearable setup where you could turn freely, but had to walk with a gamepad. So far in my experiments, being able to freely turn with a wearable VR setup is a big factor in adding immersion. Using a manual treadmill would give that up. But I guess you would also get something in return. The biggest benefit to that setup would be that you could use wired peripherals, meaning the HMZ-T1 (without hacking), a powerful desktop machine, the Razer Hydra, etc. All things that are ready to go off-the-shelf. About the only custom software you would need would be to map the treadmill motion to the 'W' key, which seems trivial compared to the other stuff that has been discussed. Hmmm....

Post by **brantlew** » Fri Oct 21, 2011 7:37 pm

@Cyber: It certainly would be a tradeoff? What's more important, real turning or real walking? I don't really have any experience to judge from. I guess it depends on the application. I would be interested to know how natural the turning mechanic could become once you got used to it. Twisting your body at least resembles the real world motion versus pushing a joystick which in no way resembles walking.

Having thought a bit more about it I realize that I didn't think through a lot of the details. For one, you're right - those treadmills don't go backwards. Also they are hard to start moving and once moving they continue moving if you just stop immediately. Palmer's Curve treadmill looks like the right solution but those things are $5,000!! So for a usable system you might still need to build a custom motorized and computer driven treadmill - similar to things we have discussed elsewhere in this thread. The difference being that you only have to work in one dimension which simplifies the task a lot.

Edit: As far as mapping motion to the W. If that's the best we can get then I almost don't see the point. The reason you feel immersion from real turning is because it maps 1:1 with your body. Mapping 1:1 with the walking speed and motion should be the goal and I bet would be equally immersive.

Post by **Synexious** » Fri Oct 21, 2011 7:42 pm

A bidirectional treadmill has advantages. Manual treadmills can be had for less than $100, and they can be used to walk on backward and forward. I originally envisioned using one in the simulator I've planned, but cyber got me thinking being able to turn is better. However, people who used the GameRunner prototype said they felt very immersed. Feeling the sensation of walking normally through cyberspace is very immersive. Of course, so is turning. It may be we have to choose one or the other, realistically. Perhaps turning in place would be better for games like Call of Duty and using a treadmill would be better for The Elder Scrolls V: Skyrim. FPSes require twitch play, so turning happens alot more and alot faster. RPGs are slower and involve more walking over long distances, so I think a treadmill might be better in that case. And with a standard treadmill wired devices can be used. It's a complete plug and play solution. It doesn't get any cheaper or easier.

Rowan Underwood goes jogging - in Cyrodiil! For serious running, though, I don't think you can really use an HMD. You'd need a monitor or projector.
[youtube-hd]http://www.youtube.com/watch?v=_H-HrQKsUF4[/youtube-hd]

How does this ODT work? Do the belts bend? It looks different from other ODTs with wider belts moving perpendicular to the chain.

[youtube]http://www.youtube.com/watch?v=msci440q18s[/youtube]

Post by **PalmerTech** » Fri Oct 21, 2011 7:52 pm

I have seen that, and I think that it does not work.

Watch the video closely, 90% of the walking is in only two directions. He never travels more than a few feet in other direction, and always turns back to one of his primary directions very quickly, or backs up. I think he knows very well the limitations of the design, and that video is made to try and hide them.

EDIT: Never mind, looks like I am wrong. Pretty suspicious video, though! Check out here: http://www.vsd.bz/" onclick="window.open(this.href);return false;

EDIT AGAIN: Okay, maybe I am right. No explanation of how it works beyond that single diagram, no pictures or video of behind the scenes, and that video he posted on youtube still has the limitations I outlined.

Post by **brantlew** » Fri Oct 21, 2011 8:07 pm

@Aphradonis: I agree. I think slower paced games like SkyRim would definitely benefit more from the walking mechanic.

I was actually thinking today of a hybrid solution for Second Life where you would jump on a treadmill to explore but then be able to hop off and freely move within a small 8x8 square when you are just hanging out in one place. Admittedly would suck for most games but for purely academic VR its "ok".

Post by **Synexious** » Fri Oct 21, 2011 8:19 pm

I like that idea, Brant. Palmer, someone in the comments said it's a room moving over a floor. Why would anyone even bother doing that?

Post by **brantlew** » Fri Oct 21, 2011 8:48 pm

Check this out. I guess that's how you solve the problem of starting the treadmill motion without handles or inclines. Now just need to figure out how to not fly off the back when you stop. Maybe another flexi-cord on the front?

[youtube]http://www.youtube.com/watch?v=aZyU5ZJV5pE[/youtube]

Post by **Okta** » Fri Oct 21, 2011 9:01 pm

The treadmill theme would be worth a try but doesn't really touch on 360 head tracking which really is the core of the immersion. I think even running on the spot would have it beat.

Post by **brantlew** » Fri Oct 21, 2011 9:21 pm

Okta wrote:The treadmill theme would be worth a try but doesn't really touch on 360 head tracking which really is the core of the immersion. I think even running on the spot would have it beat.

I guess I am thinking in terms of independent head and torso rotation. Head movement is pretty much taken care of with all the existing tracker hardware. The treadmill does not address rotation of the torso, so you would need a physical controller or gesture system for that.

Confession: I admit I'm pretty ignorant of the capabilities of existing games in terms of programmable controls - so I am just assuming best case scenarios where you have hooks into all positional variables.

Post by **Synexious** » Fri Oct 21, 2011 11:33 pm

A standard treadmill with an HMD would combine the ability to look (but not turn) freely with natural walking.

Post by **mAchiNE** » Sat Oct 22, 2011 5:07 am

brantlew wrote:I like mAchiNE's idea of using lower profile bearings and sinking them into thick soled shoes to minimize the height impact. The second unit looks more realistic to me since the first one is so wide that you would not be able to fit many together. These are all really large bearings (1" or so). At that size we could only get a few of them on there and they would really be more like rollers skates. Have you guys run across any smaller size bearings - maybe in the 1/2" range?

http://newzealand.rs-online.com/web/p/b ... s/0687821/" onclick="window.open(this.href);return false;
15mm ball bearings in this assembly (0.59")

Post by **pierreye** » Sat Oct 22, 2011 10:56 am

I actually aiming for this.
http://malaysia.rs-online.com/web/p/products/0687635/" onclick="window.open(this.href);return false;

Just need to cut off the bolt fitting. The problem with flange unit is you can't mount the ball transfer unit close enough due to the fixing centres = 29mm. Bolt fitting is cheaper too. I estimate I need 50 units for the whole project and come close to USD 300. Now I'm thinking if I can go cheap first to try out the concept by getting a bowl shape fiberglass, put some baby powder on it to make the surface slippery and wear socks.

Or use a children sandbox such as this (Diameter close to 4') or 4' satellite dish.

http://www.amazon.com/Step-2-740500-Cra ... 388&sr=1-6" onclick="window.open(this.href);return false;

Post by **cybereality** » Sat Oct 22, 2011 12:51 pm

Not sure if this has been mentioned or not, but is there any reason an off-the-shelf pair of roller-blades could not be used?

You would need a harness, and probably some sort of curved bowl to walk on, but it seems like a really affordable and no-nonsense solution.

Post by **PalmerTech** » Sat Oct 22, 2011 12:55 pm

Hard to turn. Omnidirectional shoes would probably have the front tip of the shoe made of a rubber pad, so you could stand still and turn easily.

Post by **cybereality** » Sat Oct 22, 2011 1:08 pm

Its possible to turn with roller-blades. Obviously harder than wearing sneakers, but I think it would be even easier than with the omni-directional ball bearings we are talking about. And you could probably get brakes installed on the roller-blades (front or back) pretty easy, and its all off-the-shelf stuff.

Post by **mAchiNE** » Sat Oct 22, 2011 4:03 pm

yes but you can't walk sideways in rollerblades and stay in the same place like you can with ballbearing shoes

Post by **cybereality** » Sat Oct 22, 2011 5:29 pm

@mAchiNE: Well, yeah. Walking sideways would not be possible, but that seems secondary to forward and backwards motion.

Just trying to come up with something that is realistic. Dunno about you, but I certainly don't have $5,000 to spend on a cyber-carpet.

Post by **Okta** » Sat Oct 22, 2011 6:09 pm

How about these instead

Post by **cadcoke5** » Sat Oct 22, 2011 6:41 pm

Regarding all of the VR related motion devices, I have always had a doubt about the ability to navigate it easily. My doubt is because most of them can't simulate the user's inertia. E.g., when you stop, a person will actually lean back for a short time to compensate for his sudden deceleration. You make similar compensation when you start moving or turn. Furthermore, you experience the forces on your feet.

The "Hamster ball" VR environment;

The only exception I can think of to my concern over inertia, is the giant hamster ball. In that, when you stop running or walking, the ball tends to continue rolling, so the user must compensate in a similar way to stopping on flat ground. If the ball's rotational inertia is equal to the forward inertia of the user, then it is a build-in solution. But, when the user has stopped the ball's rotation, then the user is beyond a neutral position. The ball would roll back, and then over-shoot the neutral position. The user would have to compensate by leaning forward. The real world doesn't work like that. However, perhaps the system automatically applies brakes to stop the ball from that roll-back situation. Regardless of the method, the demonstrations I have seen of the hamster ball show that users are able to use it. The main problem is that this is not something to home-brew.

Roller shoes;

For the roller shoes discussed here, I suspect that they won't be viable. I have a little experience walking on ice, and it is a difficult thing to do. I suspect the harness will be carrying your weight most of the time. I think the experience would be like using your feet like a computer mouse.

=========================================
My own suggestion on a hackable VR environment

I have always thought it would be easiest to simulate the user on some sort of scooter. Another user here suggested a jet-pack. Car and plane simulators are another example, but perhaps they are more removed from the environment than something like a 4-wheel ATV.

I picture the user seated on an ATV or scooter of some sort. Then, this hanging by a cable from the ceiling. Actuators can push the ATV in any direction to allow the user to feel as though they are accelerating or stopping. The final touch would be for an actuator to simulate bumps in the road by pulling on the cables the ATV is hanging from.

Joe Dunfee

Post by **WiredEarp** » Sat Oct 22, 2011 7:21 pm

@ Okta: thats actually a pretty good idea. those shoes are already low profile and have rollers!

@ cadcoke: I know what you mean about the inertia effects, but the question remains if they are really necessary to provide a realistic sense of walking. I suspect that the motion of walking is much more important to this sensation than the inertia on stopping.

Actually, regarding this as well, while its good and nice to aim for a 1:1 realism mapping between reality and VR, this may not be required, or even wanted, by the vast amount of users. I've been surprised how easy it has been to get used to doing 180 turns to turn 360 (with both my gun and helmet providing turning input) and in fact, as long as you dont NEED to map to the real world, its more fun to use and still JUST AS IMMERSIVE, if not more so (probably due to the head movement). I think PalmerTech hinted at this earlier when talking about his stage setup at his work. He mentioned that many people preferred 'accelerated' movement.

I suspect that once we all start playing VR games etc, 'accelerated' movement may actually become more popular than real world movement. After all, given a world we can make our own, who wouldn't want ot be able to jump further, run faster, turn quicker?

Post by **brantlew** » Sat Oct 22, 2011 9:10 pm

The last couple of posts about the importance/unimportance of inertial effects got me thinking about another novel idea. This is definitely not on a DIY scale, but it seems near perfect as an institutional VR room for a single user. Imagine an empty rectangular room - say 100ft long and 50ft wide. The entire ceiling is decked out with optical sensors for accurate full body tracking. The user is only equipped with HMD equipment. All movement is natural and unimpeded. This is similar to the facilities that Palmer has described before.

The cool thing about this room however is that you can walk uninterrupted in a straight line forever in it. Here's how. The floor consists of two giant computer controlled rotating circular pads - each one 50ft in diameter and near touching at one point. As you move outward toward the edge of a circle, it will gradually rotate so that as you step off the edge of one circle, you will be stepping onto the edge of the opposite circle. The floor basically does it's best to keep you always traveling towards the center of the room and if successful, you will never hit a wall.

Here's an example. The colored lines represent the path that the user takes along his journey. The user starts in the center of the bottom circle and starts walking northward along the straight blue line. As soon as he crosses onto the top circle it begins a slow 180 rotation so that by the time he reach the center he is facing south. Now he will follow the green path on his southward march back to the starting point.

Here is another example of a user that starts out on an eastward trek. (EDIT: Oops. Just realized that the top part of this path should be curved just like picture above. You get the idea though.)

As long as the circles are large enough and the acceleration graduated the user will feel no angular momentum. That combined with the immersive visuals will convince the user that he is walking in a straight line. This could go on indefinitely. There is no limit to the size of the virtual world. It just has to be flat. In practice there are some problem areas in the fulcrums between the circles that might have to be dealt with, but in principle the user could make any sequence of turns and movements and never hit a wall.

Am I crazy? Am I missing something crucial here or would this work as I have suggested?

Post by **cybereality** » Sat Oct 22, 2011 9:25 pm

@brantlew: Thats an interesting idea. I feel like it may have issues if the user were to strafe while he was near the connecting point. But if it were mostly linear motion, I think it could work. The question would be if the user would feel the circular motion, and if this would ruin the simulation.

Post by **WiredEarp** » Sat Oct 22, 2011 9:50 pm

Haha thats a pretty cool idea. Be fun to play with additional effects you could provide by having a moveable floor

Post by **brantlew** » Sat Oct 22, 2011 9:57 pm

I believe it's just a matter of scale and speed. You have to keep the radius large enough so that the circles can accelerate at an imperceptible rate for the time it takes for the user to cross one radius. The faster the allowed speed, the larger the circles need to be. I picked 50ft diameters out of thin air just thinking in terms of walking speed. To handle fast running would probably require much larger circles but the principles are the same.

Really the worst part of this design is the large size required - but in terms of complexity this is actually one of the simplest mechanical designs that I have seen. Mechanically, it could very well be a DIY project. It's just that nobody has a gymnasium size space to work with.

Post by **brantlew** » Sat Oct 22, 2011 10:15 pm

@Cyber: The examples given use simple linear paths to illustrate the idea, but I believe that any change in direction and path could be handled. Viewed from overhead, the path would look really complicated. It also gets tricky right at the junction between circles if the user decides to do a 90 degree turn at that exact moment. Then they will be traveling out-of-bounds in the fulcrum. The one thing that helps though is that the distances are so large that they have ample opportunity to accidentally step back into the circles, and once they do that their path can be corrected. I guess as a back-up mechanism you could put triangular treadmills in the fulcrums that would try to guide the user back to the center (just like the Battlefield 3 simulator floor we saw earlier)