| Author | Message | ||
     Anonymous Rating: N/A |
Analog devices now has gyroscopes in IC form. An Eval board can be purchased for 50.00 combined with a dual axis accerlorometer (also from Analog for a few dollars). If you can get together a Hardware and software guy you could produce a REALLY cheap 3DOF tracker. Another piece of the puzzle now available to the consumer. -Kevin | ||
| BOPrey Rating: N/A |
Do you have a link to the source where I can buy a development kid. I can build the hardware and write the driver for it for Win2K. One thing about these accerlorometers though, they tend to drift over time, which screw up the tracking. My favor is the ones using ultrasound technology from Logic Tech. | ||
| Anonymous Rating: N/A |
Sourceless, non-line of sight is the key I believe to effective trackers. I agree drift will defintely be a key issue, but I think it can be done. Basically what we would be building is a cheaper consumer level Intertrax 2. There is no reason at all that a 3DOF tracker should cost 900 dollars. I can build the hardware no problem. Just need a software guy. You can go to Analog's website www.analog.com and look under gyroscope. -Kevin | ||
| BOPrey Rating: N/A |
Kevin, I have Intertrax 2. Aside from the money, it works very good. Maybe I can help you on the software side, and can be talk about this through private messages? | ||
| Anonymous Rating: N/A |
You would be better off with a Gyro Mouse you can buy at BestBuy. | ||
| BOPrey Rating: N/A |
I have a GyroMouse too. It is not that good to use as head tracker. | ||
| Anonymous Rating: N/A |
Yeah I agree but it uses the same technology.. | ||
| Anonymous Rating: N/A |
Gyromouse is a 2DOF sensor. We would be incorporating a gyro from analog for Yaw measurement only and then use accelorometers for pitch and tilt. You could use the logic of "uses the same technology" for any tracker ever built. BOPrey, I'll be contacting your shortly once I do some more research. Thanks -Kevin | ||
| Anonymous Rating: N/A |
I also have the Intertrax2 and am very happy with it so far. Higher latency and the fact it is for a medical application is a major factor as to why I didn't go with a lower cost magnetic alternative. | ||
| Anonymous Rating: N/A |
You can find the product under 'MEMS - gyroscopes' on the analog devices site (analog.com). There's a purchase link on the left side. I'm checking this out now to see how feasible the whole idea is. I'm building a camera-based 6DOF tracker at the moment, but I'm thinking that this would have less latency than that. - Glen Murphy http://glenmurphy.com/ | ||
| Anonymous Rating: N/A |
Err .. the gyros under MEMS - gyroscopes are only 1DOF trackers, silly me. Sorry everyone. I'll get back to looking for the ones Kevin was talking about. - Glen | ||
| tj Rating: N/A |
Hey guys... I don't mean to be discouraging, but I think the difficulty (and expense) lies not in getting the components (accelerometers, gyros) cheaply, but in the engineering. Processing the raw output of the sensors requires some complex algorithms - mainly calculating integrals (get out your calculus book! :-) - in attempt to arrive at reliable & accurate position & orientation. Unfortunately, at low velocities, the output signals are somewhat noisy, which when integrated result in errors that accumulate over time and cause drift. Next to jitter, drift can really kill the effectiveness of your tracker... to the point where you start thinking of the hundreds extra you wish you'd spent to have a tracker that works. That's why the InterTrax2 uses 3 micro-gyros, 2 accelerometers, and 2 Earth's magnetic field sensors (if I'm remembering that right)... all continually checked against one another... all just for 3DOF. ...and it still drifts sometimes! Apparently InterSense's even more accurate InertiaCube uses 3 rate gyros and 3 linear accelerometers (and no magnetic sensors I guess?)... again, all just for 3DOF. I want 6DOF... and cheap! Last year I talked to the InterSense guys at a trade show. Essentially I asked them this: "If you're using accelerometers, why can't you make a sourceless 6DOF tracker?" The simple answer is that the sensors just aren't accurate enough and there's too much error... enough to make the tracker unreliably worthless. So here's what I think you should do: Go for the visual approach, using head-mounted cameras, and from the get-go, design it to be 6DOF. What's encouraging is when you can see software (like Sense8, I believe... World Toolkit?) that can take a live feed from a camera, recognize a specific pattern printed on a piece of paper (a certain black & white shape), and overlay a virtual 3D object on that shape - so that as you move the camera, the object appearing on the screen looks to be firmly affixed on the pattern, in 3D space. This definitely seems like it could be used to do 6DOF tracking. You'd need multiple tiny cameras, maybe with wide-angle (fixed focus?) lenses... and perhaps those printed patterns could be stuck to the ceiling. Then someday in the future, hopefully the system could be made to recognize its surroundings in lieu of the patterns. | ||
| BOPrey Rating: N/A |
ti, I think using ultrasound for 6DOF tracking is much much cheaper than camaras. InterSence and LogicTech both make these type of tracking devices. The only cons is speed, it could only do a max of 50 samples per second (largely due to the velocity of sound trevaling in the air), and it is dead accuate. The one from LogicTech sells for about 2 grand. I think Kevin can make it much cheaper than that. | ||
| BOPrey Rating: N/A |
Oh, I almost forgot. ti, If you want to fix you drift problem with the mechanical gyro, use a laser one. This is how they work. http://www.fas.org/man/dod-101/navy/docs/fun/rlg.htm | ||
| Anonymous Rating: N/A |
Camera based tracking can be dead cheap. If you're just after head tracking in a sitting position, all you need is a webcam and 3 LEDs (or more if you want a wider range of movement) This can get you highly accurate, zero-drift results. It's just a bit (~300ms) laggy. -Glen | ||
| BOPrey Rating: N/A |
Glen, Good idea. BTW. I don't thing 3 LEDs will give you 6DOF though, and you have not talked about external interfearence, such as light level in the evironment. Oh, 300ms of lag is not acceptable for any kind of vr application. | ||
| tj Rating: N/A |
BOPrey: If you want to build a cheap sonic tracker, maybe the Power Glove is the way to go. You can get 'em cheap enough on eBay that you wouldn't have to feel guilty for tearing one up to reconfigure for HMD use (like you might if it cost $2000)! I don't know though, maybe it's not accurate or fast enough... but perhaps you could tweak it. I have one (a Power Glove, not torn up) and I was always pretty impressed with how well it worked versus its cost, but I didn't do any kind of real analysis of it. Are you doing something like that, Kevin? I wonder if you have to worry about any Mattel patents! BOPrey: That laser gyro sounds interesting, but is it available? If so, I wonder about cost. If it costs too much, a slew of cheaper components might still be preferable (and let's face it, money IS eventually an issue for everybody... unless you're military, I guess)! Glen: I like the 3 LED idea! I'd think that you COULD do 6DOF. The camera-detected distance between the dots could potentially tell you just as much as sonic "time-of-flight" delays. It's all triangulation. If all 3 LEDs were identical, I'm not sure if the system could get confused or not (maybe not if the dots were CONTINUALLY tracked by your software after calibration... hmmm, I'll hafta keep thinking about that one). If that would be a problem, perhaps different LED colors or filters could be used or maybe they could be pulsed in some way (if it didn't affect latency). As far as lag from the camera goes, couldn't better CCDs be used? There are higher-speed CCD-based cameras available, capable of recording slow motion -- like special purpose ones used for machine vision (e.g., detecting faults on an assembly line). However, there again, it might get down to cost... which I haven't looked into at all. I'm just shootin' from the hip here! Oh, and as long as we're throwing out ideas... As an alternative to the 3 LEDs, I wonder if an image could just be projected onto a ceiling? You'd have this little "projector-like" device sitting on top of your monitor. It could project 3 dots (using 3 laser diodes?), or a grid (I guess from a lamp through a transparent slide), or better yet, the gridlines could be "coded" like UPC barcodes, so that the camera could be more sure of where it was looking. If folks would be annoyed by seeing a pattern on their ceiling, use IR. Thus you wouldn't have to connect anything to your ceiling and it would be a much simpler solution that could perhaps accomplish the same thing as the grids of ceiling-mounted receivers required by high-end 6DOF trackers. I'm still looking forward to someday when we have relatively cheap, sourceless, 6DOF intertial tracking. Sourceless will be important so that it can be portable, like for AR. Keep dreaming, eh? ;-) | ||
| Glen Murphy Rating: N/A |
FWIW, I have 2DOF tracking done with one LED ( http://riot.com.au/systems.php?s=b1 ). This code can differentiate between two colours, so using an array of various LEDs, you could easily get 3DOF. You can avoid interference from outside light by installing an IR-pass filter on the webcam, and using IR LEDs (like what the TrackIR guys are doing), but then you don't get colour. I think the latency issues can be solved by doing some of it in hardware (again, like the TrackIR guys). I've seen an Augemented Reality Tracker that was a wide-angle camera in the ceiling that did 6DOF tracking for multiple people in the room, and it could cover a fairly big room, too. But I can't find much information on it. Currently, my plan is to do 6DOF with 3 LEDs and a webcam (you can't actually calculate angles without knowing position, so 3DOF isn't any easier than 6DOF). But then again, talk is cheap, so I'll shut up until I get something. - Glen http://glenmurphy.com/ | ||
| BOPrey Rating: N/A |
ti, As far as the laser gyro is concerned, no, I have not found a vender that sells one at an affortable price. Then again, I have not spent much time looking for it due to the fact that you can build one very easily with parts from either RadioShark or other electronic sources. If you are interested, I can sketch up a drawing for you. Basically, these are the stuff you need... One Infrared diode One Optical bean spliter One Fiberoptic Ring (For demostration, just use hard cored plastic tube) One Optical mixer One Infrared detector plus some caps, resisters and a battery. This will give you one degree of freedom | ||
| Anonymous Rating: N/A |
I have a gyromouse used for headtracking. It works good, but drifts up or down depending on the speed of the glancing motion. I am writing some software to compensate for this (will post a web page later). Before I bought this, I read somewhere that headtracking was available in Flashpoint and AVP2. Does anyone know how to do it? Flashpoint has a 'look around' feature, but it is tied to the mouse. I tried to map it to the joystick, but no go. (BTW, I have also written software to adapt the Gyromouse to work as a joystick.) | ||
| Anonymous Rating: N/A |
Gyromouse will work better then any other consumer level headtracker. | ||
| Anonymous Rating: N/A |
Gyromouse sucks! Mine drifts and will not stay centered. Don't waste your money on that crap. | ||
| Anonymous Rating: N/A |
If yours is drifting then you have a bad one. Buy another one, the price of two will still be cheaper then any other tracker. | ||
| Anonymous Rating: N/A |
Its actually very sensitive and its sensing you shaking. If you place it on a table and activate the gyrotracking it wont drift at all!! If you'd stop tweaking, it would stop moving (...what you call drifting. It's you who are actually drifting). | ||
| Anonymous Rating: N/A |
I am not shaking. I got two of them and they both drift and will not stay centered. CRAP! | ||
| Anonymous Rating: N/A |
I got one too and I have a solution (although a rather crazy one.) I wrote some code to compensate for the centering problem. Basically it creates a 'pit' in the center of the screen where the mouse pointer will get 'sucked in' if it gets close. When it hits the center, it pushes the middle mouse button so that the game can center too. This syncs game and mouse. If you watch the video of TrackIR in LockOn you will notice a little jarring 'snap' to center from time to time. This is what my program does. The problem with it is that you need a second computer (pentium based) and a serial cable (null cable) to make it work. The cable is only worth $10, but the computer... who knows. You can download the software here http://members.shaw.ca/drdowning/ I just made that webpage, which is why it is so poor. IGOR is a second program that will translate the GYROMOUSE output from the second computer into joystick input. I thought at first that this would have been the way to go, but LOCK ON's joystick look-around is terribe. This setup works well with Lock On and GREAT with Strike Fighters. Adjusting the mouse speed on the second computer is critical too. Drop me an e-mail if you have any questions! | ||
| Anonymous Rating: N/A |
Almost forgot! My Gyromouse goes into 'sleep' mode after 5 min's of non-clicking activity. Does anyone know of a hack around this? | ||
| Anonymous Rating: N/A |
use the newer usb gyromouse and it wont go into sleep mode. | ||
| Simon Lockwood Rating: N/A |
I am interested in developing a 2d mouse using gyros, but I am a designer and not an engineer. The use of gyros for this is to avoid the problems with optical sensing. If anyone is interested in this project please mail me to speak further. | ||
| Buckfalfa Rating: N/A |
I recently picked up a gyromouse pro for headtracking and have been pretty impressed with the results (though, as stated above drift is a problem.) I've read some stuff on magnetic interference being the big problem drift with gyroscopes- anyone have any thoughts on reducing magnetic interference? Also, has anyone had any luck taking their gyromouse pro apart? I've found that the two bug left- and right-click buttons pry up easily without breaking (if you take the battery off the back and pry loose the one square plasic piece that attaches both buttons.) I'm a bit nervous about stripping out the optical stuff, but I'd really like to try for weight reduction. I would also like to find a way to relocate the wireless mouse's battery so I don't have to wear that weight on my head either! My current solution for correcting drift in games is just to occasionally level your head and hit the "center look" button. A bit tedious, but a small price to pay for dirt-cheap headtracking. | ||
| Anonymous Rating: N/A |
Has anyone deconstructed what the X800 does with headtracking? I've heard that they've got a pretty good 3DoF tracker but no specifics. I just homebrewed a decent HMD using a gyromouse and a cheap pair of GVD300-N glasses. Immersive but not stereographic (yet- perhaps I can pipe 2 vid signals into the 2 screens?) $150 all told, and a pretty fun little project! Ran around in DOOM3 for a bit and really felt immersed. | ||
| layman Rating: N/A |
As a layman it would appear that a mechanical or optical solution (using paper patterns and a webcam) would be a more affordable 3D tracker. A mechanical 2DOF solution on first mention seems backward, but there would be no lag, drift, or electromagnetic interference. Those are pretty big advantages. | ||
| Anonymous Rating: N/A |
The TrackIR is a very affordable optical tracker, and the IR-4 tracks a 3-point object for improved accuracy. I'd love to engineer a solution similar to the way optical mice track movement, only at a variable distance. I figure that's probably impossible, though. |
