Gyroscopes and odometery?

rwgast_logicdesign

I just read phils paper on encoders and it has seriously made me change my whole outlook on odometery and how important it is. While the paper is about encoders the thing just kept making me think of ways I could implement other sensors in ways i hadnt thought of before to help out with odometery. This lead me to grow a little curious about a sensor id always dismissed in the past, the gyroscope.

Now from what i know, a gyro is used to find the angle and its rate of change, of whatever its attached to. Now obviously this is helpful when balancing a two wheel bot. I can see it also being helpful to detect when a bot is going up or down a hill, which could be used to change motor speed. But is there any other practical applications for a gyro as far as odometery goes? It just doesnt seem like its to important in a bot with more than two wheels, but at the same time gyros are everywhere, heck theres even one on my new arm board so i feel like maybe im missing something?

Ps, if your reading this Thank you Phil for one of the most helpful texts on encoder/odometery ive ever seen, and thank you Martin H for telling me about it, not a real easy to find paper if your just searching for encoder info.

Don M

Maybe you could post a link to Phils paper so that we know what you've just read?

Martin_H

Don M wrote: »

Maybe you could post a link to Phils paper so that we know what you've just read?

It's in the downloads and resources section of the wheel encoders product page.

http://www.parallax.com/StoreSearchResults/tabid/768/List/0/SortField/4/ProductID/80/Default.aspx?txtSearch=Wheel+encoders+

The sample code is worth reading through and I've reused it several times.

rwgast_logicdesign wrote: »

This lead me to grow a little curious about a sensor id always dismissed in the past, the gyroscope.

A friend of mine took an online course in building a self driving car. It described how to integrate the outputs of many sensors to pinpoint position. I don't know if it talks about inertial navigation, but it would seem a logical topic.

http://www.udacity.com/overview/Course/cs373/CourseRev/apr2012

prof_braino

Don M wrote: »

Maybe you could post a link to Phils paper so that we know what you've just read?

http://www.parallax.com/Portals/0/Downloads/docs/prod/datast/ApplyEncoder.pdf

...is the link to the paper, I think.

rwgast_logicdesign

Sorry I should have posted the link, thank you Braino for taking care of it. Its just a really good paper rich in theory.

So I was reading a page which I cant find now where I guy was using a 1500 serial based unit with a 3axis compass/Gyro/Accelerometer and some DSP filtering to do inertial navigation, which allowed him to use odometery outdoors over very long distances with little error. Its really impressive what can be achieved but the fact of the matter is the guy was using a 1500 dollar industrial sensor box, can these kind of feets be achieved with some cheap ebay/parallax sensors?

photomankc

Thus far I have found that the calculations required to get that kind of data out of the combined gyro/accel/compass modules to be demanding. I believe there is an object for the Propeller that can produce similar results but your looking at using up almost half your cogs to get it at a fast update rate. I have had decent luck getting basic info from the sensors with median filtering vs something like Kalman filtering. I have not spent any time trying to integrate those results yet because my indoor bot will not need them. That $1500 IMU is doing all the hard work most likely and the user gets to get on with navigation rather than hitting the math text books.

Something like this is a less expensive alternative:
http://www.pololu.com/catalog/product/1256

I'll likely go that route myself for an outdoor bot.

Tom C

@rwgast,
Everything you ever wanted to know about using IMUs, compasses, and odometry for a robot: http://www.geology.smu.edu/dpa-www/robo/jbot/index.html
Just a thought.
Regards,
TCIII