desktop computer on robot?

[s0crates]

My team had the idea to add a desktop computer onto the robot for additional computing power. Ignoring potential weight or power problems, would it be legal to interface it solely with the digital i/o board and other custom circuits?

-jonathan

[usbcd36]

Any fans used would need to be KOP fans or powered by KOP motors. Furthermore, you couldn't use a hard drive, since it would have a non-KOP motor. SSD or CF card in an adapter would work, though.

Assuming the cost fits within the $400 limit per part, I think it would be legal.

[s0crates]

OK, thx!

-jonathan

[EricH]

And it couldn't affect anything directly. Section 8.3.8, <R03>, <R44>, <R46-F>.

So you could use it, provided the conditions already stated and the other rules are met. It would be a custom circuit.

[Jared Russell]

You would also need to power it from the 12V battery.

That said, you won't find many sub-$400 computers that are durable enough to put on a robot yet offer a significant upgrade over the cRIO.

[Daniel_LaFleur]

Originally Posted by s0crates

My team had the idea to add a desktop computer onto the robot for additional computing power. Ignoring potential weight or power problems, would it be legal to interface it solely with the digital i/o board and other custom circuits?

-jonathan

If you are interested in going this direction you might want to look at a Gumstix.

By FIRST rules, it can read sensors, and do calculations, but cannot directly control any outputs ... the cRIO must do that. Read the sections that EricH pointed out very carefully.

[EHaskins]

As long as the cRio controllers all physical outputs there is nothing illegal about what you want to do.

Originally Posted by Abwehr

You would also need to power it from the 12V battery.

That said, you won't find many sub-$400 computers that are durable enough to put on a robot yet offer a significant upgrade over the cRIO.

If were going to do this I would probably use components similar to this:

Small 6v-26v DC-DC PSU $50 - http://www.mini-box.com/picoPSU-60-WI?sc=8&category=981
Mini ITX motherboard w/dual core 1.6ghz atom $130 - http://www.newegg.com/Product/Produc...82E16813153131
A 2gb RAM stick about $20
A 2gb to 16gb flash drive for OS and storage $10 to $50

Total price: $210 to $260

You would have to put a different cooling fan on the motherboard, but I might be temped to just run it without a fan for the match.

I would also have a duplicate storage device in case the first became corrupt due to improper shutdowns.

[viper110110]

Originally Posted by Abwehr

You would also need to power it from the 12V battery.

That said, you won't find many sub-$400 computers that are durable enough to put on a robot yet offer a significant upgrade over the cRIO.

you could always buy sub-$400 computer parts...

[s0crates]

Originally Posted by viper110110

you could always buy sub-$400 computer parts...

yup - not many motherboards cost $400+.

actually, we had found the components suggested by ehaskins a bit before his reply... Our only problem: money.

-jonathan

[Russ Beavis]

I've been pushing folks to use this for a while now and no one seems to take me seriously. Maybe it's the name? It would be very interesting to put a simple Ethernet switch on the cRIO/Axis camera and allow it to be used as a co-processor (rules permitting, of course).

http://www.cherrypal.com/SHOP.html

Russ

[Kevin Sevcik]

Sadly, the rules inexplicably don't permit ethernet switches or anything but the one KoP Axis camera to connect to the second ethernet port.

<R66> Inputs to custom circuits can be connected only to the following sources:
A. Power Distribution Board protected 12Vdc outputs
B. Speed controller outputs,
C. Relay module outputs,
D. Analog Breakout outputs,
E. Digital Sidecar PWM Out, I2C, Relay or Digital I/O ports,
F. Other custom circuits, or
G. Switches, potentiometers, accelerometers, sensors, and other additional permitted electronics.
<R67> All outputs from sensors, custom circuits and additional electronics shall connect to only the following:
A. Other custom circuits, or
B. PWM Out, I2C, Relay or Digital I/O ports on the Digital Sidecar, or
C. Analog In ports on the Analog Breakout.
D. Ethernet Port 2 on the cRIO Mobile Device Controller (to which the Kit Of Parts-provided camera, and only that camera, may be connected).

Note the ruling on the ethernet port. Note also that the serial port isn't mentioned, so that isn't an option either.** So while it's legal to use a second PC to increase your processing power, you're actually going to have an even harder time talking to it than you would have in 2008. The Gumstix would be the easiest interface, as you can get one of the expansion boards with an SPI or I2C interface, which would make things marginally less painful.

--- Nevermind the below, per Joe Ross ---
**This is confusing, as the WPI Libraries have rather good support for the Serial port, so forbidding its use seems... silly. I'm posting this to the Q&A to see if it's an oversight.

[Joe Ross]

Originally Posted by Kevin Sevcik

I'm posting this to the Q&A to see if it's an oversight.

Not an oversight: http://forums.usfirst.org/showthread.php?t=11176

[Kevin Sevcik]

Originally Posted by Joe Ross

Not an oversight: http://forums.usfirst.org/showthread.php?t=11176

This is disappointing to say the least. It's frankly ridiculous that there's support for the serial port in the FRC libraries, but we're forbidden to use it. I'm dismayed at the thought of NI spending any time on development of useless functionality when they could have been fleshing out support for actual useful interfaces, like buffered communication on the SPI bus.

There really seems to be some sort of fundamental disconnect between the 2009 GDC and the GDC of the past several years. I was under the impression that we were being given a new control system so that we could push the envelope with it and do interesting, exciting things. Instead, we're being restricted from doing anything beyond the rather strict vision of what the GDC can be done with the controller.

[JamesByrne]

Originally Posted by Kevin Sevcik

This is disappointing to say the least. It's frankly ridiculous that there's support for the serial port in the FRC libraries, but we're forbidden to use it. I'm dismayed at the thought of NI spending any time on development of useless functionality when they could have been fleshing out support for actual useful interfaces, like buffered communication on the SPI bus.

There really seems to be some sort of fundamental disconnect between the 2009 GDC and the GDC of the past several years. I was under the impression that we were being given a new control system so that we could push the envelope with it and do interesting, exciting things. Instead, we're being restricted from doing anything beyond the rather strict vision of what the GDC can be done with the controller.

I think that this is true it is disappointing that we can not do these things with the 2009 control system. But I have to agree with the GDC on this one to be honest. The control system is new and I think that there will be alot more features next year, but they really want to get all the bugs possible worked out this year. They do not want the extra complication of the possibility of custom circuits or custom computers on board, to get in the way of them making the control system (or troubleshooting it, if there is a problem).

Also the CRIO is a powerful controller I do not know why you would really need more then that right now.

Just my ideas.

[nathanww]

Going slightly off-topic:

The GDC's job is to make a good game, and then ensure that it is played in a safe and fair manner be revising rules and issuing advisories. Most of the restricitons they've placed on the control system(like the non-use of break/coast switching, the serial port, or additional network hardware doesn't fall into these categories, since none of this stuff is going to be able to override and emergency stop command in the FPGA if an emergency were to occur, and it can be regulated under all of the existing rules for budget, COTS items, etc.

I'm fine if the GDC wants to say that they won't offer support for problems in certain functions--but finding and repairing bugs is the TEAM'S responsibility, and this Apple-like attitude really seems contrary to FIRST's spirit of innovation