Douglas E Knapp wrote:I am making a device to track what a gun is pointing at. I need to be able to read hand vibrations and gross movements. I can assume a starting position that is pointing at the target and then want to track changes until the gun is fired. What is not clear to me is, should I use a 1g or 3g acc and for the gyros, what sensitivity is best? Also I keep hearing about the build up of errors but no one ever says how fast that becomes a problem. Are we talking 1/100 of a second or minutes or maybe even hours?
Also, why use an acc and a gyro? Why not just use 2 sets of accelerometers at fixed distance apart and know orientation to each other? For example a 3 axis device at one end of the barrel and another at the other end? Would this not be just as good as the acc and gyro combo?
Thanks all!
Douglas
Let me answer your questions first. I think, given what you'd mount to a firearm, that you could expect at least 30 seconds before the drift got too bad. It will depend on how well you zero out the biases beforehand and, of course, your part selection. And just how accurate an estimate of pointing you need (I'm guessing that's what you're after)
People use both because they measure different things and the fusion of their data can, if done properly, lead to a better result than one or the other alone. In your case I'd be a little worried that the relatively small accelerations would be "lost in the noise" (unless the shooter is a lot shakier than I am
) Also measured accelerations would result from both translation (up/down, right/left) as well as rotations of the gun. You'd have to know the geometry of the situation to try to separate them from each other. A gyro (in theory) doesn't care about the translations, it only responds to the rotations. Of course any translational motion would be lost with just gyros. With both you can remove the accelerations due to rotations (in theory if you know the geometry) and get the accelerations due to just translational motion.
For your application my 1'st thought is you'd get more info from a gyro package than a tri-axial set of accelerometers. You wouldn't get pointing accuracy but you would get a measure of steadiness, flinch, trigger control, etc. In theory I guess 2 sets of accel's could do the trick as well but again you'd have to know the geometry precisely and you'd now have 6 biases to null out. FWIW all of the above would only get you (at best) a pointing relative to the starting pointing. Any error in that initial position results, directly, in an error in your final pointing estimate.
May I ask ... if you're interesting in measuring pointing accuracy (again my guess @ your desired end result) why not measure it directly via some optical system. Just one way would be to mount a standard laser aiming aid and watch the target with a camera of some sort. Tracking the dot would be easy. A small camera mounted to the gun (I've seen it done w/shotguns) would work too. I'm sure there are other ways as well.
BTW would you be using this for handguns or long guns or both ?
EDIT : I kept thinking "Knapp" and shooting and it finally came to me that that's Tom Knapp. Any relation ?