computer on the robot

[whosdadog]

i want to know the rules about putting a computer on the robot. i know you can but i need to know the limitations and rules on it. any help would be great

[,4lex S.]

This is why you need a digital copy of the rules. These can be found on http://www.usfirst.org. Search in the robot section, and see what you find. Everything related to your particular application should be available in there.

If you have a specific application for your computer that is not covered explicitly, do you mind clarifying?

PS: Trying to control the robot with your computer in any way is pretty much gonna be illegal.

[EricH]

Originally Posted by whosdadog

i want to know the rules about putting a computer on the robot. i know you can but i need to know the limitations and rules on it. any help would be great

A computer would be classified as a custom circuit. That's the rule type you need to look at.

<R03> gives the interference/direct output control prohibition.
It can't cause you to go over the cost limits in <R22>.
<R40> says that it has to run off the primary robot battery.
See the rest of the wiring rules for how you should wire it to be in compliance.
<R50> says that you can't alter power pathways.
<R52> and <R53> prohibit any part with a motor--like a non-solid-state memory or hard drive. Fans would have to be KOP under the same sort of reasoning. At least, that's how the identical rule has been interpreted in the past.
<R68> governs what it can output to.

That should get you started.

[Bigcheese]

It's actually rather easy to do it within the rules. We had it setup in 2008 but never ended up using it. And now it's even easier because you can connect over Ethernet to the cRIO.

The main issues are:

• You need a DC to DC power suply that runs off 12 volts. The one we used was about $50.
• You cannot use a hard drive. Back in 2008 SSDs were still too expensive, so I simply hooked up a 8GB SD card and ran Linux.
• You have to modify the CPU heat sink to fit the KOP fans on, or use a _huge_ (and heavy) passive heat sink.
• Somehow find a solution to the BIOS battery. In 2008 the inspector knew about it and said it was ok, but I'm not sure it should have been.

[apalrd]

Do you really need another computer on your robot? The cRio contains a 400mhz PowerPC processor that all of your code runs on, and an FPGA for hardware PWM output, counters, and other IO. That should be more than enough power for anything reasonable. In 2008 and previous, the processor was a small PIC that occasionally ran out of space and lacked hardware floating-point capability, making it difficult to do trig or other complex math operations without using lookup tables. That's when it would have been useful to have a co-processor, not now.

[semiferger]

Originally Posted by Bigcheese

It's actually rather easy to do it within the rules. We had it setup in 2008 but never ended up using it. And now it's even easier because you can connect over Ethernet to the cRIO.

We try to connect over ethernet to the cRIO but CPU heat sink not modified to fit the KOP fans on.please help me.

[Bigcheese]

Originally Posted by apalrd

Do you really need another computer on your robot? The cRio contains a 400mhz PowerPC processor that all of your code runs on, and an FPGA for hardware PWM output, counters, and other IO. That should be more than enough power for anything reasonable. In 2008 and previous, the processor was a small PIC that occasionally ran out of space and lacked hardware floating-point capability, making it difficult to do trig or other complex math operations without using lookup tables. That's when it would have been useful to have a co-processor, not now.

You obviously haven't tried to do any advanced and/or low latency image processing.

Originally Posted by semiferger

We try to connect over ethernet to the cRIO but CPU heat sink not modified to fit the KOP fans on.please help me.

I'm confused as to how the CPU heat sink would effect Ethernet connectivity, but if that really is the problem, just drill some holes or glue or tape the fan on.

[David Brinza]

Check out Team 39's target tracking. They built a computer from a small single board processor, memory, etc. Per their BOM, no component was more than $200 and the total system was ~$500. The image processing and gyro integration makes for a very impressive capability!

[apalrd]

Originally Posted by Bigcheese

You obviously haven't tried to do any advanced and/or low latency image processing.

The cRio can handle processing the target this year fast enough. The only time it needs to be faster is when the robot is moving, which unfortunately leads to camera image blur. Even a faster co-processor won't fix that.

As for advanced image processing in competition, the code given to us adequately handles detection of the goal. Nothing else is designed to be tracked by the camera, making tracking difficult. You could want to track the white line to go through the tunnel, but simply seeing the camera image would help alot there.

I would agree that the cRio is not the best processor for image handling, as it is already busy controlling the rest of the robot. However, FIRST gave it to us with the intent that we would use it for image processing, and many teams have proven it is up to the task.

[Bigcheese]

Originally Posted by apalrd

The cRio can handle processing the target this year fast enough. The only time it needs to be faster is when the robot is moving, which unfortunately leads to camera image blur. Even a faster co-processor won't fix that.

As for advanced image processing in competition, the code given to us adequately handles detection of the goal. Nothing else is designed to be tracked by the camera, making tracking difficult. You could want to track the white line to go through the tunnel, but simply seeing the camera image would help alot there.

I would agree that the cRio is not the best processor for image handling, as it is already busy controlling the rest of the robot. However, FIRST gave it to us with the intent that we would use it for image processing, and many teams have proven it is up to the task.

Actually, using Weiner Deconvolution combined with the PSF generated from a gyro and accelerometer, it's not that hard to remove the motion blur. Although it's much easier to simply lower the exposure time (which is not possible to reliably do with the 206).

While yes, tracking the VISION TARGET this year is rather easy, there are more advanced algorithms that would allow more reliable and distant detection. There is also the possibility of tracking other things such as the balls (which is surprisingly easy to do on the cRIO) and other robots.

I hope FIRST keeps the rules regarding interfacing a PC with the cRIO the same in the upcoming season. We've got some custom hardware at my new job that I'd love to get running on a FIRST bot. I think the students could do a lot with stereo vision...

[theprgramerdude]

Does anyone know what altering power pathways implys?

[EricH]

The power pathway is the route the power takes to get to the motor. It's supposed to be Power Distribution Board, power control device, actuation device (or to other components as called for). If you take any other route, like bypassing the power control device, that alters the power pathway.