High speed, High accuracy compass?
Archiver
Posts: 46,084
Hi everyone,
I'm looking for a high-speed (30 Hz sampling rate or higher) chip-
based compass to use on an autonomous robot. I'd like to use the
compass for a stamp-based controller to maintain high-speed, straight-
line motion with tank-steering or ackerman-steering robots going over
15 mph down a hall. Additionally, I'd like to use the same system
for orientation control of high-speed R/C autonomous aircraft, R/C
motorcycles (got one working but it wanders), and even possibly
closed-loop control of model rockets (am I emphasizing *speed*
enough?).
I need about 2-3 degrees of accuracy, but IF the sampling rate is
high enough, I can use the stamp (or a DSP which the stamp is
attached to) to filter it.
Things I've tried that didn't work:
- Using several Dinsmore compasses in a sequential sampling routine,
i.e. reading 4 of them in order and keeping track of the results.
Unfortunately, the rise time on the sensor means that you any faster
respose with 4 than you do with just 1...
- Using the vector 2X in high-speed sampling mode. This has worked
the best so far, but I cannot get any measurements faster than 5 Hz.
Didn't bother trying to use more compasses since the same rise time
problems would likely occur as with the Dinsmore compasses.
Things that I am thinking about, but haven't tried yet. Has anyone
else tried these?
- Integrating a compass with a yaw-rate sensor (spin sensor). I'd
use the dynamic equations of motion (hard to find and even harder to
put in fixed-point stamp) to create an observer to estimate what my
heading is 'in-between' compass readings using the spin sensor.
Problem: creating multi-rate observers is REALLY difficult,
especially since I don't have dynamic models for aircraft, rockets,
or high-speed robots!
- Using a yaw-rate (spin) sensor and integrating the result to get
angle. Use a periodic compass (Vector) to bias the measurement back
to the correct angle whenever that is measured. I've seen this done,
but the robot was quite 'jerky' and the integrator drifts really
quickly (less than half a second). However: I haven't used one of
higher quality chip-based spin sensors (has anyone tried this with a
Tokin? Is that how the Roswell flyer does it?). Has anyone
integrated GPS with a spin sensor to get orientation angle?
- Gyroscopes: I bought one normally used on R/C helicopter
stabilization. The gyroscpe drift is small enough to work with, but
the price, weight, and power requirements tend to be too high (can't
put a gyro in a model rocket or on a R/C plane)
- Building my own using MEMs? (just kidding... I wish. See price
constraints below!)
Price range: hopefully less than $1,000, but ideally less than
$200. I've put double that much into trying to solve this problem
already. I don't it to cost so much that I am afraid to test it. [noparse]:)[/noparse]
Anyone out there with ANY ideas?
Thanks in advance,
- Sean,
snbrenna@u..., mr-roboto@i...
http://mr-roboto.me.uiuc.edu/snbrenna
I'm looking for a high-speed (30 Hz sampling rate or higher) chip-
based compass to use on an autonomous robot. I'd like to use the
compass for a stamp-based controller to maintain high-speed, straight-
line motion with tank-steering or ackerman-steering robots going over
15 mph down a hall. Additionally, I'd like to use the same system
for orientation control of high-speed R/C autonomous aircraft, R/C
motorcycles (got one working but it wanders), and even possibly
closed-loop control of model rockets (am I emphasizing *speed*
enough?).
I need about 2-3 degrees of accuracy, but IF the sampling rate is
high enough, I can use the stamp (or a DSP which the stamp is
attached to) to filter it.
Things I've tried that didn't work:
- Using several Dinsmore compasses in a sequential sampling routine,
i.e. reading 4 of them in order and keeping track of the results.
Unfortunately, the rise time on the sensor means that you any faster
respose with 4 than you do with just 1...
- Using the vector 2X in high-speed sampling mode. This has worked
the best so far, but I cannot get any measurements faster than 5 Hz.
Didn't bother trying to use more compasses since the same rise time
problems would likely occur as with the Dinsmore compasses.
Things that I am thinking about, but haven't tried yet. Has anyone
else tried these?
- Integrating a compass with a yaw-rate sensor (spin sensor). I'd
use the dynamic equations of motion (hard to find and even harder to
put in fixed-point stamp) to create an observer to estimate what my
heading is 'in-between' compass readings using the spin sensor.
Problem: creating multi-rate observers is REALLY difficult,
especially since I don't have dynamic models for aircraft, rockets,
or high-speed robots!
- Using a yaw-rate (spin) sensor and integrating the result to get
angle. Use a periodic compass (Vector) to bias the measurement back
to the correct angle whenever that is measured. I've seen this done,
but the robot was quite 'jerky' and the integrator drifts really
quickly (less than half a second). However: I haven't used one of
higher quality chip-based spin sensors (has anyone tried this with a
Tokin? Is that how the Roswell flyer does it?). Has anyone
integrated GPS with a spin sensor to get orientation angle?
- Gyroscopes: I bought one normally used on R/C helicopter
stabilization. The gyroscpe drift is small enough to work with, but
the price, weight, and power requirements tend to be too high (can't
put a gyro in a model rocket or on a R/C plane)
- Building my own using MEMs? (just kidding... I wish. See price
constraints below!)
Price range: hopefully less than $1,000, but ideally less than
$200. I've put double that much into trying to solve this problem
already. I don't it to cost so much that I am afraid to test it. [noparse]:)[/noparse]
Anyone out there with ANY ideas?
Thanks in advance,
- Sean,
snbrenna@u..., mr-roboto@i...
http://mr-roboto.me.uiuc.edu/snbrenna
Comments
hobbys. I use one for my electric helicopter, power is low, and size was
small also.
At 04:42 PM 4/11/2001 -0000, you wrote:
>Hi everyone,
>
>I'm looking for a high-speed (30 Hz sampling rate or higher) chip-
>based compass to use on an autonomous robot. I'd like to use the
>compass for a stamp-based controller to maintain high-speed, straight-
>line motion with tank-steering or ackerman-steering robots going over
>15 mph down a hall. Additionally, I'd like to use the same system
>for orientation control of high-speed R/C autonomous aircraft, R/C
>motorcycles (got one working but it wanders), and even possibly
>closed-loop control of model rockets (am I emphasizing *speed*
>enough?).
>
>I need about 2-3 degrees of accuracy, but IF the sampling rate is
>high enough, I can use the stamp (or a DSP which the stamp is
>attached to) to filter it.
>
>Things I've tried that didn't work:
>
>- Using several Dinsmore compasses in a sequential sampling routine,
>i.e. reading 4 of them in order and keeping track of the results.
>Unfortunately, the rise time on the sensor means that you any faster
>respose with 4 than you do with just 1...
>
>- Using the vector 2X in high-speed sampling mode. This has worked
>the best so far, but I cannot get any measurements faster than 5 Hz.
>Didn't bother trying to use more compasses since the same rise time
>problems would likely occur as with the Dinsmore compasses.
>
>Things that I am thinking about, but haven't tried yet. Has anyone
>else tried these?
>
>- Integrating a compass with a yaw-rate sensor (spin sensor). I'd
>use the dynamic equations of motion (hard to find and even harder to
>put in fixed-point stamp) to create an observer to estimate what my
>heading is 'in-between' compass readings using the spin sensor.
>Problem: creating multi-rate observers is REALLY difficult,
>especially since I don't have dynamic models for aircraft, rockets,
>or high-speed robots!
>
>- Using a yaw-rate (spin) sensor and integrating the result to get
>angle. Use a periodic compass (Vector) to bias the measurement back
>to the correct angle whenever that is measured. I've seen this done,
>but the robot was quite 'jerky' and the integrator drifts really
>quickly (less than half a second). However: I haven't used one of
>higher quality chip-based spin sensors (has anyone tried this with a
>Tokin? Is that how the Roswell flyer does it?). Has anyone
>integrated GPS with a spin sensor to get orientation angle?
>
>- Gyroscopes: I bought one normally used on R/C helicopter
>stabilization. The gyroscpe drift is small enough to work with, but
>the price, weight, and power requirements tend to be too high (can't
>put a gyro in a model rocket or on a R/C plane)
>
>- Building my own using MEMs? (just kidding... I wish. See price
>constraints below!)
>
>Price range: hopefully less than $1,000, but ideally less than
>$200. I've put double that much into trying to solve this problem
>already. I don't it to cost so much that I am afraid to test it. [noparse]:)[/noparse]
>Anyone out there with ANY ideas?
>
>Thanks in advance,
>
>- Sean,
>snbrenna@u..., mr-roboto@i...
>http://mr-roboto.me.uiuc.edu/snbrenna
>
>
>
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
Sincerely
Kerry
Admin@M...
WWW server hosting [url=Http://mntnweb.com]Http://mntnweb.com[/url]
Kerry Barlow
p.o. box 21
kirkwood ny
13795
I'm not sure a compass is going to be able to solve this problem. I think that your idea to use a yaw rate sensor will be much more successful. An angular rate Gyro such as the types that are used in R/C helicopters may be your best bet. These are piezo gyros that vibrate and can give angular rotation rates with a high degree of accuracy. This sensor would be used to maintain straight movement and the compass would be refered to occasionally for directional information to follow a heading.
I can think of many problems with a compass.· A few of which are:
1. Driving past metal objects - As your robot goes past say a large metal desk or near a car or computer or wiring in the wall, there will be interference with the earth's magnetic fields and the compass will variate from the actual magnetic heading.
2. Compass turning and acceleration errors - Compasses are subject to a variety of errors when exposed to chages in speed or turns.· Sailors and airplane pilots are well aware of these errors and that is why they primarily use a gyroscopic device to follow headings. In an airplane, the compass is only providing an accurate reading during straight and level unaccelerated flight.· The pilot occasionally re-calibrates the gyroscopic instrument to the compass during these conditions.· In the northern hemisphere, if your vehicle is on a east or west heading the compass will indicate a turn to the North when you accelerate and to the south when you slow down.· If you are on a north or south heading the compass will initially indicate a turn in the OPPOSITE direction (as much as 30 degrees opposite) in which you turn.· These are all related to compass dip and details can be found in any aviation training manual.
Original Message
From: snbrenna@uiuc.edu [noparse]/noparse][url=mailto:snbrenna@uiuc.edu]mailto:snbrenna@uiuc.edu[/url
Sent: Wednesday, April 11, 2001 9:43 AM
To: basicstamps@yahoogroups.com
Subject: [noparse][[/noparse]basicstamps] High speed, High accuracy compass?
Hi everyone,
I'm looking for a high-speed (30 Hz sampling rate or higher) chip-
based compass to use on an autonomous robot.· I'd like to use the
compass for a stamp-based controller to maintain high-speed, straight-
line motion with tank-steering or ackerman-steering robots going over
15 mph down a hall.· Additionally, I'd like to use the same system
for orientation control of high-speed R/C autonomous aircraft, R/C
motorcycles (got one working but it wanders), and even possibly
closed-loop control of model rockets (am I emphasizing *speed*
enough?).·
I need about 2-3 degrees of accuracy, but IF the sampling rate is
high enough, I can use the stamp (or a DSP which the stamp is
attached to) to filter it.·
Things I've tried that didn't work:
- Using several Dinsmore compasses in a sequential sampling routine,
i.e. reading 4 of them in order and keeping track of the results.·
Unfortunately, the rise time on the sensor means that you any faster
respose with 4 than you do with just 1...
- Using the vector 2X in high-speed sampling mode.· This has worked
the best so far, but I cannot get any measurements faster than 5 Hz.·
Didn't bother trying to use more compasses since the same rise time
problems would likely occur as with the Dinsmore compasses.
Things that I am thinking about, but haven't tried yet.· Has anyone
else tried these?
- Integrating a compass with a yaw-rate sensor (spin sensor).· I'd
use the dynamic equations of motion (hard to find and even harder to
put in fixed-point stamp) to create an observer to estimate what my
heading is 'in-between' compass readings using the spin sensor.·
Problem: creating multi-rate observers is REALLY difficult,
especially since I don't have dynamic models for aircraft, rockets,
or high-speed robots!
- Using a yaw-rate (spin) sensor and integrating the result to get
angle.· Use a periodic compass (Vector) to bias the measurement back
to the correct angle whenever that is measured.· I've seen this done,
but the robot was quite 'jerky' and the integrator drifts really
quickly (less than half a second).· However: I haven't used one of
higher quality chip-based spin sensors (has anyone tried this with a
Tokin? Is that how the Roswell flyer does it?).· Has anyone
integrated GPS with a spin sensor to get orientation angle?
- Gyroscopes:· I bought one normally used on R/C helicopter
stabilization.· The gyroscpe drift is small enough to work with, but
the price, weight, and power requirements tend to be too high (can't
put a gyro in a model rocket or on a R/C plane)
- Building my own using MEMs? (just kidding... I wish.· See price
constraints below!)
Price range:· hopefully less than $1,000, but ideally less than
$200.· I've put double that much into trying to solve this problem
already. I don't it to cost so much that I am afraid to test it. [noparse]:)[/noparse]·
Anyone out there with ANY ideas?····
Thanks in advance,
- Sean,
snbrenna@uiuc.edu, mr-roboto@ieee.org
http://mr-roboto.me.uiuc.edu/snbrenna
·
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
degree.
Also take a look in the KVH site. www.kvh.com . Maybe they have what you
are watching.
ACJacques
snbrenna@u... wrote:
>
> Hi everyone,
>
> I'm looking for a high-speed (30 Hz sampling rate or higher) chip-......