|
||
 |
Ati says | |||
| Here's an idea I had regarding force feedback gloves: what you'd do, is have a series of tubes leading from the tip of each finger to the wrist, then use an electromagnet field to draw ferrofluid into each tube to provide increased resistance. The advantages of this would be that is would be relatively easy to assemble, and fairly compact. Anyone else have any ideas on this subject? |
||||
| Total Topic Karma: 17 | - More by this Author | |||
 |
eviljawdy says |
|
||||
| I thought about stepper motors (or DC Servo's, as their cheaper) along the joint, but then this would mean your hand wouldn't be able to make a fist and all sorts of things. The ferrofluid idea sounds cool! Have you read anything about the artificial "muscles" like the contracting metals and polymers, that react under electric current? |
||||||
| - 05 February, 2007 | ||||||
|
Ati says |
|
||||
| The other problem with motors is that they tend to burn out quickly. I have read about such materials (I know quarts crystals behave like that to a limited extent). The ones with a good range of flex tend to be kind of expensive though, and I think they tend to wear out pretty easily (don't quote me on that). |
||||||
| - 05 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Ah yes, if you put too much resistance against the motors, then they will go bye-bye. I've seen some other stuff done with small holes and compressed gas, but nothing recently. |
||||||
| - 05 February, 2007 | ||||||
|
Ati says |
|
||||
| The three main problems with most rigs I've seen seems to be as follows: * They are expensive * they are bulky * They don't provide arm resistance (you can feel the cup, but you can't feel it's weight when you lift it up). Solve those and your a millionare. |
||||||
| - 06 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| * Steal it * Drill holes in bits * erm... Those were just joke answers - Agreed, they are they "grail" of realistic force feedback techniques. |
||||||
| - 06 February, 2007 | ||||||
|
Ati says |
|
||||
| Okay, that was worth a karma. One idea I had for the arm thing would be to try to use low-ampage electric current to stop the muscles from extending all the way (thus providing low-mass force-feedback. Obviously the biggest problem with this would be that the body probably wouldn't react well to that kind of current. |
||||||
| - 06 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| From what little I've actually seen regarding external electrical current to the bodies internals, haven't been received too _well_ by the body. I've not seen anything relating to adding the current to the muscles, but that seems plausable. But surely some direct link to the nerves would be of greater advantage in this kind of example/idea? Prof Kevin Warwick (University of Reading, Dept of Cybernetics - the other place that did my course :@) had some implants done to his nerves, for scientific research, and connected them in some way to an advanced robotic arm - with tactile sensors. I could imagine this being the peak of the overall idea... but seeing as how I have little to no knowledge in this area, I'm just speculating  |
||||||
| - 06 February, 2007 | ||||||
|
Ati says |
|
||||
| Oh yeah, I read about that... he had a chip that interfaced with his major motor nerves implanted in his arm. Then I think he set it up so that the computer would interpret the electric impulses of the arm into motion instructions. The really cool part is that he also reversed the connection to send tactile data as well (as you mentioned). Here's the website: http://www.kevinwarwick.com/Cyborg2.htm Now if you could figure out how to send data of that kind without using an invasive interface, you've got yourself a heck of a VR system. |
||||||
| - 06 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Just how responsive are nerves to external stimuli? How much voltage/current would one need in order to have them react? Of course, I'm trying to think of ideas that would use some sort of surface based interface, but now knowing about the specifics of bio-mechanics and neural impulses would be a major flaw in any ideas I would come up with I suppose you could have just one type of invasive procedure done at the spine, as that's where all the nerves go anyway. |
||||||
| - 06 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, nerves respond to fairly low amounts of current. I've done a few 'experiments' (screwing around with low voltages is fun), and they'll react to a 9 standard volt pulse at one amp. The real problem isn't the amount of voltage neccesary, which is negligable, but focusing it enough to effect only the nerves you want. The real problem, of course, with any proceedures at the spine is that if you do screw up the nerves that moderate your heart go through there too . |
||||||
| - 06 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Ah yes, all the normal internal organs and what not - that would be a bad thing indeed! If you encased an arm in a "unit" and this would pinpoint an impulse to a specific location (kind of the inverse of an MRI) from a 3D coordinate, would that then be possible to cause the required nerve to respond? I mean, if you had a unit that surrounded the body, or a specific area (like an arm), and this unit had thousands of impulse firing devices (kind of like a magnatron, but with very low power per device), which could choose exact regions of the limb to fire upon - would that cause the nerve to fire? Or just burn it? Or would that be power dependant? Again, apologies for some dumb questions - intrigued, and just don't know! |
||||||
| - 06 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, magnetrons would cause the nerves to fire, but they were designed for the express purpose of creating heat in organic object, and I'd really hate to see what that would do to your arm. (It's a good idea, but impractical). Not to mention that building a bunch of tiny magnetrons would be a pain in the rear. An easier way would be to find the firing threshold of the nerves, and then set your electrodes to output about sixty percent of this threshold. Then its a relatively simple matter of running several pairs of streams of this current through the flesh, and having each pair them interect on the major sensory nerve groups. This way, you could pinpoint your stimulus to an individual nerve without effecting any inadvertantly. After this, the only problems remaining are getting enough small electrodes, and decoding the neural patterns used to represent touch. And don't worry, your questions are far from stupid. I myself am running on only a slightly above average knowledge of biochemistry, and a few experiments I've done on electric current and human nerves. |
||||||
| - 06 February, 2007 | ||||||
|
Ati says |
|
||||
| It's kind of interesting to think of all this in terms of hardware- the professor on the cyborg project got a mod-chip, and we're trying to figure out how to hook up a bluetooth port. | ||||||
| - 07 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| That analogy is definately worth a Karma! I see what you mean with the Magnatron - It was the closest thing I could think of relating to the sort of idea I was having. These electrodes, would they be on the surface? Then having to go through the skin... how much current loss is there in the warm, salty water environment of humen flesh? |
||||||
| - 07 February, 2007 | ||||||
|
Ati says |
|
||||
| Yes, in my conception of this you'd use surface electrodes. There is sole current lost to skin resistance, but if you soak the skin in someting electrically conductive beforehand, it decreases enough to be easily compensated for The main nerve that handles sensory input for the hand and forwarm goes through the base of the elbow (which causes the 'funnybone' effect). So it would really only be a matter of using a 'listener' electrode to listen in on the patterns that represent the tactile sensations of various materials, then playing them back via a stimulater electrode when neccesary. |
||||||
| - 07 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| How much of this kind of technology exists? How much would need to be created? What would be the financial implications? As far as the code goes, that's relatively simple - some sort of embedded system (PIC MC would be easiest for a proof of concept) with Analogue and Digital I/O and something to "Interact" with both virtually and physically for the vict-erm-subject. In case you haven't guessed, I'm very interested and excited! |
||||||
| - 07 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, there haven't been very many commercial applications for this sort of thing (I *think* there was a gaming product a few years back that used electric current to stimulate the inner ear though). Primarily due to the negative stigma attached to running electric current through the body. There have been some scientific experiments with stuff like this. The cyborg project above was one of them. There was another project in which a man who had had both of his arms destroyed had had an artificial arm attached. The interesting part was that he had the sensors on the arm wired to nerves in his chest, and his brain 're-wired' itself to interpret impulses on the chest nerves as impulses from the hand. There have also been numerous studies done on using low voltage current to alter the way the nerves perceive sentsations (notably pain). An example can be seen below. http://www.emedicine.com/pmr/topic206.htm As you can see, there is a large base of research to work off of. The independant research that would need to be done for a project such as this, is decoding the neural signals used to represent touch, and finding the conductive materials that distort the transmitted signals the least. As for financial implications, if a product such as this could be made to work safely, and a large corporate backer found, it could easily become a major commercial venture (I know a good number of people and organizations who would likely be very interested in buying it). The real difficulties would be the legal ramificatiosn of something like this. There are a plethora of lawsuits that could be assosciated with a device such as this, so you'd want to have the terms of service drafted by several very competent lawyers. |
||||||
| - 07 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Not that you've thought about this at all ;-) Legal issues asside, these other experiments you speak of, to what extend have they been completed? Is there any way to recreate this sort of experiment in a non-lab environment (read: At home)? Hmm, thinking about this, putting even the slightest electrical current into one's body could be a very bad thing, and would need to be controlled perfectly! |
||||||
| - 08 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, of the experiments I mentioned, most of them have been completed to some extend. The pain studies met with varying degrees of success, and the robotic arm project worked like a charm (http://www.newscientisttech.com/article/dn11094), although it took a long time to train and the sensors are fairly low-rez. As for doing this at home, it is entirely conceivable. Medical 'listener' electrodes are quite cheap, and the resources to use them are made available through open EEG (http://openeeg.sourceforge.net/doc/). The real trick is stimulating the nerves, which I suspect could be done using a modified speaker. As for the dangers of electric current through the body, it's not quite as dangerous as all that. Electricity is a firm beleiver in followng straight lines, so the the only area of your body that will even be effect by voltage is the space between the two electrodes. Also, when you shock your finger on a door handle, you are being shot through by many times more voltage than we'll be using at any given time. All in all, it should be quite safe if a few simple, sommon-sense safety proceedures are followed. |
||||||
| - 08 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| That prosthetic arm is phenomenal! And the idea behind it's developmeng is so intuitive, that I can see this kind of technique moving along quickly! I was wondering how they anchored the arm, but then I saw all the straps around her shoulder. I've read a great deal about the difficulty of anchoring prosthetics to humans - either it's external anchoring (which can be cubmersome) or it's internal anchoring, which means putting a material into a warm, salt environment. No matter how corrosive resistant a material is, it still degrades inside the body (titanium is what they tend to use in hip replacements). I asked a polymer engineer doctor friend of mine about plastics in the body, but he said that even those would degrade over time. I'm building an animatronic, and intend to move to a robotic arm after that - so could be interesting to try to get the force feedback elements in there too. |
||||||
| - 08 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, I provided that example just to show that research had been done on feeding artificial nervous impulses into the body. The technique they used had the difficulty of requiring extensive brain re-training before your brain starts to interpret chest impulses as arm impulses. Which makes it too difficult to be practically used for force feedback. A better bet would be to use the idea we're talking about for your project, as it would simplify the whole thing enormously, and it would allow anyone to use it, immediately, without extensive brain re-training. | ||||||
| - 08 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Interacting directly with the existing neurons - but doesn't that assume that the nerves are always in the same place on each subject? Or would that be a small calibration issue? | ||||||
| - 08 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, the nerves tend to fall close to the same place, and if you made the electrode tips a little larger, individual calibration wouldn't be too terribly difficult. | ||||||
| - 08 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| I've read a load of details on the OpenEEG site you linked, as well as joined the mailing list. So my next set of questions would be - for say, an arm, would the active electrodes be better? The way I imagine it, is that you have a sort of "sleeve" with moveable electrode positions. That way, the "unit" is a complete system, and you just move the electrodes a little here-and-there to calibrate. Also, you mentioned controlling the muscles - but can you not just control the nerves that impulse the muscles? This is in relation to giving the idea of physical force, not just sensing movement. |
||||||
| - 09 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, wait a moment. What are you thinking of it terms of this project? are you thinking of a system in which you move your arm to control a robotic arm and feel what the arm feels? Or something else? Active electrodes give you a cleaner signal, but they are more expensive, and probably more prone to error given the larger number of active components. Also, if you did want to force the real arm to move in a certain way, you COULD pinpoint the nerves, but the muscle fibers make a much more conveniant target. |
||||||
| - 09 February, 2007 | ||||||
|
Ati says |
|
||||
| Correction: I meant to say that active electrodes are more prone to CATASTROPHIC failure. To put it another way, they'll give you a cleaner signal if they work, but their more likely to stop working. |
||||||
| - 09 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| I was just thinking about some basic movement-to-robot type venture. So you move your arm, and the robot moves, then some sort of tactile response from the 'bot to the "sleeve". Of course, if that seems a bit redundant, then be sure to let me know! I figured that doing it this way would make for a faster turnaround and not put so much a requirement in good quality electrodes. And regarding active electrodes - the stuff on OpenEEG seemed like they were reasonably simple to produce, and if I utilised one of the battery operated versions, then less likely to give any kind of shock, and _should_ work. But then again, it may be worth while to use passive electrodes for the 1st iteration, and move on to some sort of active electrode "test" version at a later date. |
||||||
| - 10 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, that sound like a perfectly feasibal project (In terms of the force feeback system anyway; I have little or no experience building robotic arms). The mechanics of it depend on how you want to track hand movements; for this you really have three options: You can use an off the shelf dataglove, you can use a camera to track colored dots on your fingers, or you can include an electrode into the sleve and try to decode the instructions coming down the motor nerves. For my project, I will be using a camera simply for cost effectivenesss and full body tracking, but you may have differert reuirements. If you decide to do motor-nerve listening, it will give you more work to do, given that you'll have to decode the signals that occur when trying to bend your arm. On the other hand, it'll give you more precision, and will give the sleve a very low profile. If you use the camera it'll be cheap, and fairly low profile, but it will also not work well if the hand is tilted at certain angles relative to the camera. The off the shelf data glove will be high profile, and it'll pe spendier, but it will aso be very easy, and it will definitely work. Once this has been decided, the only thing left to do for either of us is to get to work actually assembling the system decided on. |
||||||
| - 10 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| A friend of mine made a "data glove" design, quite cheap too - using off-the-shelf bend sensors. All his theory was sound, just ran out of time, as usual in a student project. My actual plan was not to think too hard about the motion detection (yet), but more about the EEG side. As I've done object recognition and motion capture before, so I know the theory, and getting it to run in practice would be "relatively" simple (read: well documented). As for the robotic arm, the sort of thing I'm building atm, is a 6DOF basic arm out of a book. Once I understand it's intricacies, I do intend to build a humanoid arm - with more realistic features, that match the human arm. Then make the EEG device that would control this new arm. The camera/motion-capture system can readily be integrated with the ARToolkit system in some way, which would take out a lot of math calculation type things. Maybe :-S |
||||||
| - 10 February, 2007 | ||||||
|
Ati says |
|
||||
| So basically, right now your thinking more of the force-feedback elements of the project? |
||||||
| - 10 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Yeah, as the way I imagine it, that's a little easier to implement. And once in place, then something more daring and "impressive". I'm interested in this project - but it's gonna have to join the queue, what with my animatronic being built (1st) then me learning 3D game engine development (2nd) and then this... looks like I'll be pretty busy for a while |
||||||
| - 11 February, 2007 | ||||||
|
Ati says |
|
||||
| Personally, the project at the top of my queue is my VR project. This is composed of parts: 1: Create a functional VR gaming engine 2: Rig up a functional HMD 3: Get head/body tracking up and running 4. Impliment a neural interface for force feedback. So the force-feedback project is part of the the project at the top of my queue, but its the last part of the project. |
||||||
| - 11 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Seems doable - what do you mean by VR-Engine? Do you mean an entire 3D gaming engine, with some different I/O renderers and functions (like steroscopic vision etc). Also, the current wave of trackers (imho) are too expensive, not responsive enough (without spending A LOT) for what they are. But, taking off-the-shelf stuff into account - everything exists, the integration of them all doesn't, so that'll be your greatest challenge... but then, we geeks love all of that  |
||||||
| - 11 February, 2007 | ||||||
|
Ati says |
|
||||
| Well, for the gaming engine I was planning on setting up a system using the Direct X 9 shader engine, with a stereoscopy system and full motion support (i.e. finger motion, complex foot motion, etc). I was planning on using a simple camera capture system for motion tracking ( a very simple one too - you wear a different colored dot on each finger, your head, and each foot. The program counts the number of pixels in a certain color range that occur in the camera input, and uses that to get X position. Then it takes an average of all pixels in the color group to get Y/Z position. The hardest part will probably the neural interface. |
||||||
| - 11 February, 2007 | ||||||
|
eviljawdy says |
|
||||
| Interesting - the Camera based tracking idea has been used a lot. I remember a while back seeing something BT (British Telecom) had experimented with based on camera tracking, to animate an artificial "Presenter". And the ARToolkit will especially help there - at least with some of the theory and modules, if not the whole application. |
||||||
| - 12 February, 2007 | ||||||
|
Ati says |
|
||||
| Yeah, the AR toolkit probably will come in handy, for theory if nothing else. The only problem with the camera idea, is that it can't track you if you move your limb behind your body, or elsewhere where it isn't directly in the camera's view. | ||||||
| - 12 February, 2007 | ||||||
|
|
