Use of laptop on robot

[Daniel_LaFleur]

Originally Posted by popo308

just out of curiosity why do you want to put a laptop on your robot?
I could understand for dashboard stuff on the o/i but the robot? (unless you are using it for some crazy powerful camera tracking)

My team tried this a couple of years ago with the vision tetra.

We used an industrial PC board (486 based), a solid state drive w/ linux OS (small kernal, fast boot ). We used a 5v DC-DC power supply to supply the 5volts to the board (the board required both 5v and 12v). We used an evaluation copy of a LabView inspection program (available to everyone). We were able to track and pick up the vision tetra ... but it took too long to run the calculations so we never was able to score with it.

[Kevin Sevcik]

Originally Posted by Robostang 548

Thats a good point. But I suppose if a pc system did get expensive enough you could individually price each component. We got ours on ebay for less than $100 and it included the ram and processor.

-Don

Unrelated to this thread, but I'd be a little uncomfortable using a price off eBay for my cost accounting. The GDC often notes that you need to use COTS Supplier prices for things, and I'm not sure any old eBay store would count. I'm quite certain a personal auction of sale wouldn't.

I'm not saying you can't get parts from such places, but you certainly need to get a price from a more standard retailer like NewEgg or something. And if you can't find that particular system at a standard retailer, then you'd need to price individual parts and add it all up. Of course if you can't even find the parts at a standard retailer, then you're in a pretty difficult position since they wouldn't be legal. The reasoning there being similar to the reason we can't use our nice non-jittery kit joysticks from 4 years ago and instead have to use the current kit joysticks or find passable modern substitutes.

[Al Skierkiewicz]

Originally Posted by Billfred

I can't put my fingers on the Q&A at the moment (if they're even up still), but I recall the GDC historically deeming hard drives illegal as they use non-kit motors.

This is the ruling from the condensed Q&A

8.3 Robot Rules
Hard drive on robot?
Hard drive on robot?
Posted by pmm at 01/10/2007 12:54:21 pm
Are we allowed to have a hard drive on the robot for processing information? Or would we
have to use a storage device that has no motor?
Re: Hard drive on robot?
Posted by GDC at 01/11/2007 08:24:38 pm
No, as a disk drive would include an additional motor and would therefore violate Rule <R46>.
Solid state storage devices may be used, as long as they are used in compliance with all rules
regarding custom circuits.

Personally, I think that allowing this type of device might prove interesting in the long run. However, from an inspection point of view, the "custom circuit" would require some serious investigation as to mounting and securing of said device.

[Kingofl337]

This would be cool....

http://www.engadget.com/2007/12/10/v...act-barebones/

[Madison]

Originally Posted by Kevin Sevcik

The reasoning there being similar to the reason we can't use our nice non-jittery kit joysticks from 4 years ago and instead have to use the current kit joysticks or find passable modern substitutes.

I've always been under the impression that the COTS rules apply only to the robot.

[larrysz92]

depends on this years rules if your aloud or not!!

[Kevin Sevcik]

Originally Posted by M. Krass

I've always been under the impression that the COTS rules apply only to the robot.

Madison,

Under R11 from last year, the OI was excluded as part of the robot for weight, volume, and cost determination, but was expressly included in the robot definition for all other purposes, including COTS parts sourcing. However, R28 provided a slightly odd loophole, in that it gave permission to re-use unmodified robot components from previous years, even if they're no longer available, provided they weren't custom made for FIRST like the '06 transmissions, etc.

So I stand corrected in that a team that owns older joysticks that aren't modified can use them. But COTS rules apply to anything new you acquire for your OI.

Kevin

[EHaskins]

I looked into this last year. The most parctical thing I found was either a PDA(with a serial port), or a Gumstix unit.

[robogeek753]

If you want to send data to the guys at the driver station, just mess with LabView and work it like that. You can use the best laptop you can afford, and you don't have to worry about the rules. The only thing it couldn't do is help with autonomous.

Really, the vision system is way to hard to use, don't try it (3 years of failing)

Good luck if you do try though

[Binome]

I still think that the vision system is not too overly complex for FRC teams.
Your practically given a fully doccumented, sources given example find/track driver, that gives you the current size and posisition of the blob thats precalibrated for the green light.

As a rookie programmer, with absolutely no FRC controller experiance, I had the camera dumping useful information on the second day I had my hands on the controller.

Our team did not implement a successful auto mode, simply due to time contraints and lack of forethought for feedback mechanisms. This is what appears to plauge most teams that had no auto mode. It's not the hardware. The CMUcam is ideal for the scenerio that teams were given last year. The software is extremely powerful, yet abstract enough that programmers do not have to worry about manually deciphering the datastream coming out of the camera. It's precalibrated... I don't understand why people are so angry with it.

[Kevin Sevcik]

Originally Posted by Binome

I still think that the vision system is not too overly complex for FRC teams.
Your practically given a fully doccumented, sources given example find/track driver, that gives you the current size and posisition of the blob thats precalibrated for the green light.

As a rookie programmer, with absolutely no FRC controller experiance, I had the camera dumping useful information on the second day I had my hands on the controller.

Our team did not implement a successful auto mode, simply due to time contraints and lack of forethought for feedback mechanisms. This is what appears to plauge most teams that had no auto mode. It's not the hardware. The CMUcam is ideal for the scenerio that teams were given last year. The software is extremely powerful, yet abstract enough that programmers do not have to worry about manually deciphering the datastream coming out of the camera. It's precalibrated... I don't understand why people are so angry with it.

I'll agree, provided that FIRST actually gave teams a reasonable chance to actually debug during practice rounds. We had the vision system semi-working for Aim High. I say semi because on the practice field it would home on the light, aim the turret, and spin up the shooter to the proper speed to hit shots dead on. Sadly once it was on the field, it simply wouldn't recognize the light or anything. If we could have actually monitored the debug port during a practice round I'm sure we could've had it working perfectly.

[Donut]

Using the vision system successfully for autonomous is not impossible, obvious by the fact that teams did get it working. The problem is usually testing and debug time. Our team had a fully working autonomous in the 2006 season that scored balls in the high goal, yet in 2007 we weren't able to score due to missing test time. The code was better than the previous year's, and sections of it worked, but actually getting it all connected to work never happened (at the AZ regional this past year we discovered, after all our practice time was up, that our camera had been off the whole time, due to being too rushed to check for that problem).

I'm interested in seeing what exactly is going to be done in autonomous mode that would require an extra processor just to do it.

[whytheheckme]

I'm still not sure what purpose a full blown PC on the robot serves.

But...


Now I'm super interested. I'm sure you could piece together a small PC with a passive cooling system that can use a KOP fan and run solely off the battery for less than $100 a part. Solid state storage could be as simple as a flash drive that is booted from USB through the BIOS. Those are like..... practically free now-a-days...

Now just to find a reason for all of this....


Jacob

[Daniel_LaFleur]

Originally Posted by whytheheckme

I'm still not sure what purpose a full blown PC on the robot serves.

But...


Now I'm super interested. I'm sure you could piece together a small PC with a passive cooling system that can use a KOP fan and run solely off the battery for less than $100 a part. Solid state storage could be as simple as a flash drive that is booted from USB through the BIOS. Those are like..... practically free now-a-days...

Now just to find a reason for all of this....


Jacob

I can give a good reason.

When we were trying to track the vision tetra we needed a lot of computing power to to image recognition within a 3d space. This was because, while we knew the size and shape of the green panel on the vision tetra, it could be orientated and positioned so that the shape and size would be difficult to interpret.

The best PC board that we could find in the FIRST budget range, at the time, ended up being insuffiecient for our needs ... but it was a great learning experiance for the students anyways.

In the end we were able to track, move towards, and pick up the vision tetra while understanding where we were on the playing field. We just didn't have the computing power to do it and cap within the autonomous timeframe.

With todays small inexpensive industrial PC boards ... it should be possible to do.

[tajmorton]

Originally Posted by whytheheckme

Now I'm super interested. I'm sure you could piece together a small PC with a passive cooling system that can use a KOP fan and run solely off the battery for less than $100 a part. Solid state storage could be as simple as a flash drive that is booted from USB through the BIOS. Those are like..... practically free now-a-days...

Another option would be using some kind of an SBC (Single Board Computer). The TS-7200 that EmbeddedArm makes runs Linux and requires no cooling. It does cost $150, but it has a lot amount of processing power (200MHz ARM9 + 8MB RAM). It boots pretty quickly (~3secs with the stock Linux, IIRC), and has serial (TTL), SPI, and GPIOs.

Just another option...

- Taj