How to estimate roll position via gyroscope data?

[usb232]

im trying to implement a tilt sensing program via accelerometer and gyroscope. part of the input that i will plan on using is gyroscope data. since gyroscopes measure angle/sec, integrating this will give me angle right? i then have three questions:

1. do i multiply the gyroscope's (1/sampling rate) to the deg/sec to get degrees?
2. if i am to do the first one, for lack of a better term am i going to "discretize" the value? that is, any angle measurement i take is multiplied to a constant = (1/sampling rate) OR
3. do i do something like this 1/ (sampling rate*n) where n is the sample number?

thank you!

 

[serbring]

im trying to implement a tilt sensing program via accelerometer and gyroscope. part of the input that i will plan on using is gyroscope data. since gyroscopes measure angle/sec, integrating this will give me angle right? i then have three questions:

It's right

1. do i multiply the gyroscope's (1/sampling rate) to the deg/sec to get degrees?
2. if i am to do the first one, for lack of a better term am i going to "discretize" the value? that is, any angle measurement i take is multiplied to a constant = (1/sampling rate) OR
3. do i do something like this 1/ (sampling rate*n) where n is the sample number?

thank you!

take a look to this
http://en.wikipedia.org/wiki/Numerical_integration

Remember that numerical integration imply a numerical drift in steady signals

Hope this helps

 

[usb232]

i would like to begin by saying thanks!
ive read through the article but my understanding of it is not that good. I am sorry.

it says in this part that to use

rectangle rule

for integration i am to use this formula:

however it requires that i evaluate at points a and b. the only data i have is (b-a) which is delta, or the gyroscope's sampling rate. does this mean that i won't be able to estimate the angle?

 

[serbring]

i would like to begin by saying thanks!
ive read through the article but my understanding of it is not that good. I am sorry.

it says in this part that to use for integration i am to use this formula:

however it requires that i evaluate at points a and b. the only data i have is (b-a) which is delta, or the gyroscope's sampling rate. does this mean that i won't be able to estimate the angle?

Maybe I'm misunderstanding what you have written, but f(a) and f(b) are the sampled values of angle rate, so you have all data you need

 

[usb232]

sorry about that. i should have given a clear example.

1. suppose at time ta i get a converted gyro reading of 20 deg/sec. i then multiply this to (1/gyro_sampling rate) = 2seconds to get 40 degs ----> this is f(a)?
2. at time tb i get a converted reading of 30 deg/sec. i again multiply this to 2 seconds to get 60 degrees ----> f(b)?
3. what would be my (b-a)?
4. how do i calculate f ( (a+b) / 2)? i don't have the function f T_T

 

[serbring]

sorry about that. i should have given a clear example.

1. suppose at time ta i get a converted gyro reading of 20 deg/sec. i then multiply this to (1/gyro_sampling rate) = 2seconds to get 40 degs ----> this is f(a)?
2. at time tb i get a converted reading of 30 deg/sec. i again multiply this to 2 seconds to get 60 degrees ----> f(b)?
3. what would be my (b-a)?
4. how do i calculate f ( (a+b) / 2)? i don't have the function f T_T

if at t=ta you read 20 deg/s and at t=tb you read 30 deg/s, a=ta, f(a)=20, b=tb, f(b)=30.

Hope this helps

 

[usb232]

am i right in assuming that since (1/sampling rate) = delta then
ta = a = n * delta and
tb = b = (n+1) *delta
then b-a = delta too?

and f( (a+b) / 2) = (20 + 30) / 2 = 25?

 

[serbring]

am i right in assuming that since (1/sampling rate) = delta then
ta = a = n * delta and
tb = b = (n+1) *delta
then b-a = delta too?

and f( (a+b) / 2) = (20 + 30) / 2 = 25?

that's true