Example gyro code released.

[about_abot]

Originally Posted by Kevin Watson

Has anyone tried this code yet? I'd like to get a sense of how (if) it's working for teams.

-Kevin

Kevin,

First of all, thank you for your undying effort keeping this huge event on track. I just started this year as a programming mentor and do not know the ins and outs of what has happened over the last few years with the code.

The programming student on the team is new, and he is a little anxious with everything. He first saw "C" just a few weeks ago and is feeling a little overwhelmed. I am guiding him through how to layout the functionality he needs to meet the required game rules and translate all of that into working code.

In one of your previous posts, you stated you were working on a new robot.c (and I hope a new robot.h). Do you have any idea when we would see that? Having that will help immensely since my role is to help and not perform the actual work instead of the student doing it himself.

Respectfully,

Gary D.

[Kevin Watson]

Originally Posted by about_abot

In one of your previous posts, you stated you were working on a new robot.c (and I hope a new robot.h). Do you have any idea when we would see that? Having that will help immensely since my role is to help and not perform the actual work instead of the student doing it himself.

As the FIRST folks haven't sent our demobot back for us to test updates on, I'll post the changes (hopefully) sometime this weekend when my team has their 'bot moving around. If you want to make the changes yourself, only two things need to be done to get robot.c to work with the new gyro code:

1) Because I now return angular units expressed in milliradians instead of degrees, line 198 needs to be changed from:

heading = Gyro_Angle()*PI_MRAD/180L;

to:

heading = Get_Gyro_Angle();

2) The programming interface was expanded and many functions were renamed. To fix this, I would just drop the new version of gyro.c/.h into your build directory and let the compiler find the deltas for you (there will only be a few).

-Kevin

[about_abot]

Originally Posted by Kevin Watson

As the FIRST folks haven't sent our demobot back for us to test updates on, I'll post the changes (hopefully) sometime this weekend when my team has their 'bot moving around. If you want to make the changes yourself, only two things need to be done to get robot.c to work with the new gyro code:

1) Because I now return angular units expressed in milliradians instead of degrees, line 198 needs to be changed from:

heading = Gyro_Angle()*PI_MRAD/180L;

to:

heading = Get_Gyro_Angle();

2) The programming interface was expanded and many functions were renamed. To fix this, I would just drop the new version of gyro.c/.h into your build directory and let the compiler find the deltas for you (there will only be a few).

-Kevin

Great! Thank you for your help. I will let you know if we get it to work this weekend.

:-)

Gary

[Kevin Watson]

Originally Posted by about_abot

...I will let you know if we get it to work this weekend.

Cool. I have no idea if people are actually using the code I've published over the last month or so and would very much like feedback on performance, ease of use, etc...

-Kevin

[Tom Bottiglieri]

Kevin, the gyro code has worked fine for us. We are able to track heading, and stick to a heading. We arent using the scripting, for lack of need.

As soon as we get our encoders in, I will post how the PID works out.

[Kevin Watson]

Originally Posted by Tom Bottiglieri

Kevin, the gyro code has worked fine for us. We are able to track heading, and stick to a heading. We arent using the scripting, for lack of need.

As soon as we get our encoders in, I will post how the PID works out.

Very cool. Which gyro are you using? Which method are you using to calculate the gyro bias? Calc_Gyro_Bias() or the manual method I described? Are you calculating the bias once at startup or are you actively tracking it with frequent calls to Calc_Gyro_Bias()? Sorry, 'bout all the questions, but I'd like to get a sense of how much drift people are seeing from a larger sample of gyros. I have some serious magic that I can throw at the problem if people can't use their gyros because of too much drift.

-Kevin

[Tom Bottiglieri]

For now, we are using the BEI gyrochip provided in the 2003 kit. This is for testing purposes, seeing as they are illegal for robot use. We are getting gyro bias at startup, with the Calc_Gyro_Bias() function (we plan to poll it more, but we have only done basic tests, seeing how we are learning this as we go). Our gyro use code hasnt really been developed, since we are waiting for our encoder delivery, and we are planning to tackle the PID system all at once. But in our initial tests, we were seeing a bit of drift; The heading would waver about +- 17 milliradians every 8 seconds. (I dont believe we had the scale factor set the right way, seeing as one rotation was giving us a heading of about 10000 mRads.)

In order to keep costs down, I was hoping to use the provided accelerometer in place of a yaw gyro for the bot. Im guesing this would work with the gyro driver you wrote, but is there anything I need to change? (yanno.. something that I would probably never be able to figure out on my own.)

As for the drift, we have explored 2 options. One was to mount the gyro to a piece of foam tape or rubber, hoping it would cut out motor/compressor vibrations. I think the gyro bias is being set incorrectly due to the vibrations, and since for now we are only polling for bias once, it may have ill effect on our code.

Any tips? Thanks.

[Kevin Watson]

Originally Posted by Tom Bottiglieri

For now, we are using the BEI gyrochip provided in the 2003 kit. This is for testing purposes, seeing as they are illegal for robot use. We are getting gyro bias at startup, with the Calc_Gyro_Bias() function (we plan to poll it more, but we have only done basic tests, seeing how we are learning this as we go). Our gyro use code hasnt really been developed, since we are waiting for our encoder delivery, and we are planning to tackle the PID system all at once. But in our initial tests, we were seeing a bit of drift; The heading would waver about +- 17 milliradians every 8 seconds. (I dont believe we had the scale factor set the right way, seeing as one rotation was giving us a heading of about 10000 mRads.)

Yeah, you'll need to set the scaling factor before you know how much drift you're seeing. I didn't think to add the old gyrochip as one of the options in gyro.h, but will look into it.

Originally Posted by Tom Bottiglieri

In order to keep costs down, I was hoping to use the provided accelerometer in place of a yaw gyro for the bot. Im guesing this would work with the gyro driver you wrote, but is there anything I need to change? (yanno.. something that I would probably never be able to figure out on my own.)

I've already released TI accelerometer code. Here's the post. BTW, you can't use the accelerometer to detect yaw. It's great at measuring pitch or roll, though.

Originally Posted by Tom Bottiglieri

As for the drift, we have explored 2 options. One was to mount the gyro to a piece of foam tape or rubber, hoping it would cut out motor/compressor vibrations. I think the gyro bias is being set incorrectly due to the vibrations, and since for now we are only polling for bias once, it may have ill effect on our code.

Any tips? Thanks.

Yes, set the scaling factor correctly (I'll try to update gyro.h to include the gyrochip) using the instructions in readme.txt. Then try the manual method I documented in readme.txt (also in gyro.h, I think). I have a ADXRS150EB that, once correctly calibrated, drifts only a few milliradians a minute, which is pretty good.

-Kevin

[neilsonster]

For now, we are using the BEI gyrochip provided in the 2003 kit. This is for testing purposes, seeing as they are illegal for robot use.

Where did you hear they are illegal? Check the Q&A under ID 1118:

Q: Can we use Gyros from previous year's kits on our robot?
A: Yes as long as it passes the 2005 Part Use Flowchart and you account for its cost according to the cost accounting rules.

EDIT: Nevermind... I guess the Q&A people didn't look up the price...

[Tom Bottiglieri]

Originally Posted by neilsonster

There shouldn't be a problem with you using the gyrochip on your robot.

Sweet, thanks for the info. I was under the impression it was illegal seeing as it was up for sale in a FIRST yard sale.

[Kevin Watson]

Originally Posted by Tom Bottiglieri

Sweet, thanks for the info. I was under the impression it was illegal seeing as it was up for sale in a FIRST yard sale.

I've updated the gyro code to include scaling factors for the two different BEI GyroChip gyros that FIRST has included in the KOP in the past. The code can be found here: http://kevin.org/frc.

-Kevin

[CyberWolf_22]

I don't believe that the BEI gyro's are legal because the lowest price I have been able to find them for is approximately $350 which over the $200 price limit for additional electronics. If anyone has seen a gyro for sale cheaper then that, that would follow the vendor requirements that FIRST has laid out please post a link because I have 3 of them sitting in the shop that I would love to use on this years robot.

[kutty18]

Originally Posted by Kevin Watson

I have some serious magic that I can throw at the problem if people can't use their gyros because of too much drift.

Hmmm..Magic?...We tried to calibrate the gyro, but there was too much drift. We mounted it to a square piece of wood, but we still had too much drift. Is there any way to control the drift?

Thanks

[Kevin Watson]

Originally Posted by kutty18

Hmmm..Magic?...We tried to calibrate the gyro, but there was too much drift. We mounted it to a square piece of wood, but we still had too much drift. Is there any way to control the drift?

Thanks

What kind of gyro are you using? Did you try the manual method that uses Set_Gyro_Bias()? Just make sure the gyro is motionless as the bias is being calculated.

-Kevin

[kutty18]

Originally Posted by Kevin Watson

What kind of gyro are you using? Did you try the manual method that uses Set_Gyro_Bias()? Just make sure the gyro is motionless as the bias is being calculated.

-Kevin

We are using ADXRS300ABG & ADXRS300EB-ND Gyros. Yes we tried the manual method that uses Set_Gyro_Bias()?