2

i am working on an app where i need to find velocity and distance traveled without using GPS. i made some R&D on Accelerometer and motion Framework.but unable get any idea to accomplish my work.Any type of suggestions are welcome. Thanks in advance.

garry007
  • 49
  • 9
  • 1
    See also [Need to find Distance using Gyro+Accelerometer](http://stackoverflow.com/questions/6647314/need-to-find-distance-using-gyroaccelerometer/6648538#6648538) and [Getting displacement from accelerometer data with Core Motion](http://stackoverflow.com/questions/4449565/getting-displacement-from-accelerometer-data-with-core-motion) – Kay May 29 '13 at 13:52

1 Answers1

3

What about physics..

v = v0 + a * t

v0 = initial velocity, v = final velocity, a = acceleration, t = time

d = v0*t + (1/2)a(t^2)

Of course, precision will be function of polling time

AlexWien
  • 28,470
  • 6
  • 53
  • 83
Andrea
  • 26,120
  • 10
  • 85
  • 131
  • These are the equations for moving with constant acceleration. In this problem area there is no contant acceleration even within small periods. Thus it would need a [numerical integration](http://en.wikipedia.org/wiki/Numerical_integration) or better a Simpson rule based approach. Anyway this is not the main problem, but (1) the lack of an external reference and (2) the extreme error that is propagated; After 0.2 - 0.5 seconds drifting starts to exceed the actual values and signal noise ratio is too bad to get useful data. – Kay May 29 '13 at 14:38
  • Ne Smart , it's just a delta, in discrete world is a differece between two points in time, the faster is your sampling freq the more accurate is your data. Don't downvote for it. – Andrea May 29 '13 at 15:25
  • I understand you. But the main problem is, that this definitely doesn't work out that way, even when integrating with sampling at 1 MHz (100 Hz is maximum). There are still errors because gravity can never be determined correctly. It's an unsolved mathematical problem which has every statellite, rocket, car outside of GPS, ... A lot of people are asking this question over and over again. They hope for a solution but there is none. That's why a voted to close the question. Specify this clearly and I not only revoke my downvote, you'll get an upvote on top :-) – Kay May 29 '13 at 18:06
  • @Kay if the device is fix mounted (inside a vehicle), the gravity can be calibrated. Do you have a reference (e.g paper) where the problems are stated and discussed? – AlexWien May 29 '13 at 18:53
  • Yep it's a difficult topic, the main issue is that to get a correct measurement as you say is nearly impossible, because there are too many variable and, of course, you will need to use gyro, maybe also the compass. I don't know which are Garry purpose but maybe trying to simplifying as possible, forcing the use case just in one way, something better can come up into mind. We should always think that they are small embedded device. Thanks Kay, really appreciate ;-) – Andrea May 29 '13 at 18:55
  • @AlexWien s. the really good answer from Ali in [Need to find Distance using Gyro+Accelerometer](http://stackoverflow.com/questions/6647314/need-to-find-distance-using-gyroaccelerometer/6648538#6648538). You get the best results when combining all sensor data in a sensor fusion algorithm like a Kalman filter. So they kind of calibrate themselves mutually. The main problem is the different behaviour of gyro and accelerometer regarding the latency. While a gyro reacts almost instantanously, accelerometers have a latency. That means: the faster the movement the faultier the gravity estimation. – Kay May 29 '13 at 21:37
  • @Andrea I updated the answer to revoke my downvote. Was an interesting discussion :-) – Kay May 29 '13 at 21:56
  • @Kay Your edit was invalid, please add your opinion to your own answer. – AlexWien May 30 '13 at 11:50