Maker Pro
Maker Pro

Position & Orientation measurement help!

I've started my senior project and I'm running into some questions..
What the project is, is a modular device that will need to record the
x,y,z displacement from its original location and its orientation
(tilt, yaw and roll degrees). I've been searching around and it looks
like the orientation can be done with a mixture of gyro sensors and
analog output compasses (i think it's called a dinsmore compass). But
anyways, what's boggling me is a solution to calcluate the current
position. I've read about the accelerometers and it sounds like
there's a lot of error involved. And our spec is to have the device
measuring for 30mins, at a sample rate of ~30Hz, so I am assuming there
can be a lot of error introduced and propagated through time. And I
thought of GPS before we got some specs, but due to their slow reading
rate that is out of the question. Does anyone know of a good way to
measure position other than using accelerometers? or is there a way to
perhaps use a three axis accelerometer without much errors, or
filtering out errors? Thanks!
 
I've started my senior project and I'm running into some questions..
What the project is, is a modular device that will need to record the
x,y,z displacement from its original location and its orientation
(tilt, yaw and roll degrees). I've been searching around and it looks
like the orientation can be done with a mixture of gyro sensors and
analog output compasses (i think it's called a dinsmore compass). But
anyways, what's boggling me is a solution to calcluate the current
position. I've read about the accelerometers and it sounds like
there's a lot of error involved. And our spec is to have the device
measuring for 30mins, at a sample rate of ~30Hz, so I am assuming there
can be a lot of error introduced and propagated through time. And I
thought of GPS before we got some specs, but due to their slow reading
rate that is out of the question. Does anyone know of a good way to
measure position other than using accelerometers? or is there a way to
perhaps use a three axis accelerometer without much errors, or
filtering out errors? Thanks!

Search on "inertial navigation". Some of the predecessors to GPS - like
Loran - may still be around, and mobile phones have a pretty good idea
where they are with respect to the local transmitters, not that I've
got any idea how you could exploit that.
 
M

MK

Jan 1, 1970
0
I've started my senior project and I'm running into some questions..
What the project is, is a modular device that will need to record the
x,y,z displacement from its original location and its orientation
(tilt, yaw and roll degrees). I've been searching around and it looks
like the orientation can be done with a mixture of gyro sensors and
analog output compasses (i think it's called a dinsmore compass). But
anyways, what's boggling me is a solution to calcluate the current
position. I've read about the accelerometers and it sounds like
there's a lot of error involved. And our spec is to have the device
measuring for 30mins, at a sample rate of ~30Hz, so I am assuming there
can be a lot of error introduced and propagated through time. And I
thought of GPS before we got some specs, but due to their slow reading
rate that is out of the question. Does anyone know of a good way to
measure position other than using accelerometers? or is there a way to
perhaps use a three axis accelerometer without much errors, or
filtering out errors? Thanks!

From basic equations of motion:

s = (at^2)/2

so a = 2s/t^2

If you want 1m error after 30 minutes the offset error you can tolerate in
an accelerometer is

2/3.24E6 m/s/s or roughly 0.1uG.

You can buy inertial guidance packages off the shelf and one this good will
cost > $100k.

If you can think of a way of resetting your position every few seconds in
some way then the accelerometers get a lot more reasonable.

Of course there are other error sources and it gets worse if its moving
fast, tilting etc.

Look at NAV420 on www.xbow.com - this lloks like a GPS/sensor combination -
no idea how much it costs but it may give you ideas.

Have fun.

Michael Kellett

www.mkesc.co.uk
 
A

artie

Jan 1, 1970
0
I've started my senior project and I'm running into some questions..
What the project is, is a modular device that will need to record the
x,y,z displacement from its original location and its orientation
(tilt, yaw and roll degrees). I've been searching around and it looks
like the orientation can be done with a mixture of gyro sensors and
analog output compasses (i think it's called a dinsmore compass). But
anyways, what's boggling me is a solution to calcluate the current
position. I've read about the accelerometers and it sounds like
there's a lot of error involved. And our spec is to have the device
measuring for 30mins, at a sample rate of ~30Hz, so I am assuming there
can be a lot of error introduced and propagated through time. And I
thought of GPS before we got some specs, but due to their slow reading
rate that is out of the question. Does anyone know of a good way to
measure position other than using accelerometers? or is there a way to
perhaps use a three axis accelerometer without much errors, or
filtering out errors? Thanks!

What's the volume you need to cover? If the volumes are relatively
small (cubic meters), you might look at polhemus technologies
(www.polhemus.com) -- they make motion and orientation tracking
systems.
 
I'm not exactly sure on the range of volume that will be tracked. We
are still waiting for some detailed specs.. Assuming from the range of
recording of 30 minutes, it can have a great amount of displacement.
But ultimately this device is to be put on like a videocamera, so I
guess the MAX would be a person walking in one direction for 30 minuts
(1~2 miles?). I'll dig into the inertial guidance and polhemus tech
and see what I can dig up.. Thanks so far!
 
A

Adrian Jansen

Jan 1, 1970
0
I'm not exactly sure on the range of volume that will be tracked. We
are still waiting for some detailed specs.. Assuming from the range of
recording of 30 minutes, it can have a great amount of displacement.
But ultimately this device is to be put on like a videocamera, so I
guess the MAX would be a person walking in one direction for 30 minuts
(1~2 miles?). I'll dig into the inertial guidance and polhemus tech
and see what I can dig up.. Thanks so far!
You can have simple, cheap or accurate. Choose one of these attributes
at the cost of the other two.

--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.
 
Top