Shiny Doom Forum

[rklee]

Max, you had said you wanted to do a VR glove right? I just picked up my copy of Mechanical Engineering today... look at it http://www.memagazine.org/
coincidence that we both mentioned our gloved projects today? hehehe

Perhaps we can share info.
For myself I want to make a gloved keyboard. It will need to learn your typeing style first kinda like voice typing. It should be fun thopugh. I think it can be done by using 5 different inputs.
bending resistors at the three joints, distance between fingers, and accellerometers on each finger.
bending to distinguish between ujn or ikm etc.... the distance between finger mainly is for distinguishment between rt and fg and fb... etc and also accelerometer to determine when you are pressing a key.
Should be fun
any comments or suggestions?

for VR they are using pistons it looks like small pistons. What were you thinking of using? I'm very interested in your system also... would be fun (perhaps adapted for whole body for really cool VR hehehe).
oh yea I was also thinking maybe it would be easier to do this... the glove has several led's attached to each finger and there's a camera you sit on your monitor or something that tarcks your hands. that might also give some very useful data to work with.
_________________
"My Heart Is In the Work" - Andrew Carnegie

[Trool]

OOO those things were pretty cool.. I'm not sure what to use for the vr part, I'm thinking the whole thing will have to be a robot arm that can take input from the user's hand and then respond if the input is validated. But i think there might be an easier way to build one then what they had on those pics...

I'm not sure how many sensors and things i'll need, but it might be a lot, i haven't really thought about basic design yet, hopefully will soon, then i'll post things here when i think of something. I liked ur idea for a bending refference point for one finger and then finding where everything is just by that...I wana see if i can use that...only thing is, if i want to user to feel objects and stuff, then i don't know if i'll need that since the user is limited by the robotic hand and it's not that hard to know where it is at all times...so that's more design ideas that i've still got to think about...lol
I'll post more when i figure stuff out, but I wana start coding things/ordering parts before thanksgiving so hopefuly i'll have some ideas soon...
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[rklee]

so this glove isn't just cabled? there is a robotic arm on it? If so they have similar type systems with gloves. The robot arm ends with this pen and this pen is used as VR... you can poke objects and they'll react accordingly... push too hard and you can break the VR object. I think they have two of such machines in one of our robotics labs (I think feedback control systems lab) I was going to take that class next semester but I cant due to schedualing conflicts.
_________________
"My Heart Is In the Work" - Andrew Carnegie

[Trool]

Thanks for all those link...i've checked them out and now have an idea, but i need to find more info to see if it works, i'm thinking it might be possible to use muscule wire for this, but i know very very little about it. I'll try to do some more research on how fast it can contract and relax and how easy it is to control, if you know anything about it or know of good places to find info please tell me...
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[rklee]

Actually I do know quite a bit about muscle wires. I was researching it for a class last year. Muscle wire (nickle titanium I think... nitinol) is a shape memory alloy it changes shape with elecrical power. I do not think this is a good diea for your VR project for several reasons:
Muscle wire requires a lot of power... <very inefficient>
It has very little strength for how much power it requires requires multiple strands to do very much.
Contraction % is very very small. I think you can get apporx 15% on some of the better wires.
Hard to control. Like muscles they contract with voltage however unlike muscles it cannot be controled all the easily and also it is only contraction that is powered so you'd need two sets of wires to do contraction and expansion.
It's response is fast however it's relaxation is very slow.
Also it's quite expensive as compared to motors.

I think muscle wires are cool but not very practical. They've used it for small light walker robots and also those butterfly ies that run on a battery (don't actually fly). But I don't think it has enough power to do a VR glove. Consider how many strands of muscles we have in our hands.... although nitinol can generate more power per size than our human muscles... it's still quite a bit of wires.
On the other hand, they are making muscle wire pistons (I think robotstore and jameco has them) that might be better. They use a helical muscle wire... but eh lol I'm biased but I don't think there is any real practical application to this yet.
I'm not 100% sure on this but they do generate a lot of heat too don't they???
_________________
"My Heart Is In the Work" - Andrew Carnegie

[Trool]

Yah the more i was looking at it the less i liked it...it's very hard to find places that just talk about it without trying to sell u some...Things i found were that they really only contract about 3-7% witch might be enough, but they recoment u stay in 3-5% zone so that u don't burn it out...Temperature wise the low temp ones are activated at i think 52C and reach 70C and the hi-temp ones go up to 90C I think that's what i remmember...that would be very hard to build a system where the user was protected, but the wire was still being heated and cooled repidly to relax...Some of the really nice ones can pull about 3kilo worth of weight wich is impressive, but the couple sites i found that talked about working with the stuff described it as a very tricky exersise since the the whole thing is based on resistance and the length of the wire you have to be able to regulate the voltage and/or the current and ppl say that is a pain... So i don't think i'd be using that, but it would have been interesting...lol
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[Trool]

this is kinda what u wanted to do i think...i can't find the video demo i saw on techTV but the website is: http://www.lightglove.com/ It's not very good...the thing uses reflections from an led to capture key strokes and mouse movements...the demo was pretty impressive...
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[James]

I was busy learning about Nitinol for one of my classes (Biomaterials 2). Name comes because it's a nickel-titanium alloy, and it was first made in the Naval Ordinance Laboratory. Ni-Ti-Nol. And all the stuff Reg said about it is true, which is why it probably won't be used to replace muscles in the body. Just felt like sharing.

[rklee]

The light glove looks like a good idea but I still find it lacking. I don't see how it can be very accurate unless it's software is very very good or the typist is very well trained. I believe checking bending is a much better method for determining where the fingers are if it is to be used similar to the way our traditional typing works.
I believe if we were to reinvent typing... their method might be interesting... but I believe it is too simplistic to only measure 10 distances. I think this also limits the versitility it also. It would be much more versitile to know exactly where the fingers are. We can get more exact information for relating to preset templates for typing. I think it would be very nice to build a learning program that can set up templates for each person that can be set up in a file. I don't think this would be too difficult.
Detection of the key storkes can be done by locating when he finger's z axis is changing directions. This is the point where we would measure the 3 bending resistance values. We can just have a program tell the user to type out several documents. I think the hardest part is to set up the punctuations.
I dunno I still think it isn't too hard to do the regular letters. Just make a program that detects the values each time you type the letter a. Make some sort of histogram and weight it's probablility dependent on that.

Max - I'm still not sure what you're trying to build.. is it a glove that's only wired? so the position of the hands are not tracked? or is it attached to some robotic arm so not only are the relative distances on the hands are tracked but also the global position is tracked.
_________________
"My Heart Is In the Work" - Andrew Carnegie

[Trool]

what i wana build is a system that can track the locations of every finger in real space not relative to the hand...also i'd like to be able to track every joint, but that might be to annoying and complex...i also want to give some kinda feedback to the hand depending on what type of object they might be squeezing, but that i'm not totaly sure off...

i'd like to find a better way of doing all of this then attaching a bot arm to the human arm, but i can't really think of anything that could do the same things...i'm thinking of ditching the feedback idea for now and just work on iding where the finger ends are in real space

Just a slight more tidbit on the lightglove, i'm not sure how the whole thing works totaly, but the guy originaly designed it for his mom who has arthritis and he made it so that it would adjust to just about anyone, i think his goal to to provide an alternative typing for ppl who can't use normal keyboards....
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[Trool]

On of the ways that i'm thinking this might work is if i seperate this into two systems, one to track the motions of the hand and the second to stop or counteract those motions...

For the second part i'm thinking of making a semi arm with free moving joints, but be able to control how much friction is in the joint, hence how hard it is to bend it. I'm not sure what this type of joint is called, it's kinda like a sphere with something that can rotate freely around it, the stick that's doing the rotating has a shaft on the inside of it that can press on the ball, depending on how hard it presses that will generate more friction and hopefuly enough to stop the joint from bending, I think u can control the force of the pressing shaft with that coil idia, i think it's called selinoid, but don't remmember well, it's when u coil wire around a shaft and when u pump current through it it will move one way or the other through the shaft....

On major issue i can see with this idea is that wear is going to be big, but maybe if i don't make it too forceful the wair might not be too much of a problem... what u think of this...? can u suggest a better way of controlling the power needed to bend a joint, i'd like to give it as much freedom as possible....?

hope my explanation helps, i'll try to make a picture of this thing i'm thinking of maybe that will help more...lol

Thanks for help
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov

[rklee]

From a Mech E point of view... that approach is not very good. You listed wear... that's a major point however I think the hardest thing is making it ... That's not an easy thing to make... sphereical joints are difficult as it is... but adding some sort of friction to it is hard. Also... Friction is incredibly difficult to control. So stuff like exercise bikes can use friction but that's manually adjusted... it's tough to get any type of correlation between normal force and friction force. especially since it's a sphereical joint... I think it's going to be difficult... I'll have to think of a good suggestion before I suggest it
_________________
"My Heart Is In the Work" - Andrew Carnegie

[Trool]

thnx for the help, the only other thing that i can think of is maybe using servoes to control the joints, but i think that will be big and very bulky, also it will be more of a responce to the user input rather then naturaly just monitoring it and interfering only when necessary. I think that's y i liked the joints idea better, though with the servoes i will know exactly the possition of each joint so i wouldn't need to have a sensor glove type thingie, so that does have benifits....though i don't know of how to make it so that the bulky servoes rn't interfering and making everything huge and unmanagable...
though the nice thing is that most joints only really bend in one direction so there wouldn't be much need for multi directional ones....
ohh well still more thinking needs to be done....
_________________
The most exciting phrase to hear in science, the one that heralds new discoveries, isn't 'Eureka!' but rather 'hmm....that's funny.'

--Isaac Asimov