Linux-Powered Robots and Flexible Programming
 

http://www.chiefdelphi.com/forums/sh...1&page=4&pp=15

This is a continuation of the "Let's Have Linux Robots" thread, which I thought would now appeal to the Programming folks a lot more.

After some summer hacking fun, I got a set of proof-of-concept scripts together. From the release notes:

If you're interested in trying it (requires a Linux laptop), see http://adambots.gotdns.com/cgi-bin/v...rceCoProcessor for an overview, and http://adambots.gotdns.com/cgi-bin/v...onPreviewNotes for getting started directions.

Re: Linux-Powered Robots and Flexible Programming
 

Wow. Many have thought about this, you actually did it. Awesome!

Re: Linux-Powered Robots and Flexible Programming
 

I actually blogged about this some time ago. I'm glad to see someone's tried it!

Re: Linux-Powered Robots and Flexible Programming
 

Well... now I just have to get a USB/serial convertor which actually works... :D

Nice work. I will try this out as soon as I can get my hands on an old robot.

Re: Linux-Powered Robots and Flexible Programming
 

Looks impressive. I'll be very interested to see how updated versions pan out.

Re: Linux-Powered Robots and Flexible Programming
 

Regarding power, we've successfully powered our embedded chip of choice (Gumstix + CF + Serial) with a 12V-5V DC regulator, which should be allowed in the rules (harmless, 500mA max) as custom electronics. Worse comes to worse, we'll use the 5V from a few digital IO pins, lol.

Re: Linux-Powered Robots and Flexible Programming
 

Hmm... If you install Apache and use some CGI, you could make the robot controllable through a web based interface. Although, Java would probably get you better real-time performance.

Props on the project though.

Re: Linux-Powered Robots and Flexible Programming
 

I think mod_perl would get you the max performance, since it caches and precompiles. Python with psyco would get you interesting performance too.

And since the final solution (http://ubuntubackports.org/adambots-...modular/README) will be TCP/IP based to begin with, you can use WiFi to control the robot remotely while developing your robot software... Imagine that, tweaking the robot without lugging it back within reach of the serial cable, pushing the annoying button that's next to the sharp pointy pins that I've stabbed myself with multiple times.....

Re: Linux-Powered Robots and Flexible Programming
 

I like this but I'm wondering if it will be legal for the FIRST games. This is probably something that should go to FIRST's FAQ (I don't think they had this in mind when they wrote those rules!). I hope it is legal it would allow for some interesting robots! :D However I also feel this pang that makes me wonder if this gives an unfair advantage over teams unable to do this (assuming its legal :roll eyes: ). Our school has a JAVA class and if you can get the robot to function on JAVA our team will probably have 30-40 able programmers! (I'm serious!) Most programmers at our school don't do it in robotics because its not object oriented and were too lazy to do anything but java! :D

Re: Linux-Powered Robots and Flexible Programming
 

I think as long as the co-processor is under $200 this is perfectly legal. By that logic you could also say that teams shouldn't build arms because it gives teams an unfair advantage over those who don't have the resources to build an arm. If you have the resources and time, it should be perfectly legal. A rule that would make that illegal would pretty much halt the innovative think tank that FIRST has become.

Re: Linux-Powered Robots and Flexible Programming
 

It only differs from the coprocessors that many teams have been using for years in that it is more powerful.

So i suppose we should have a rule that allows inspectors to reject mechanisms because they are too sophisticated?

Re: Linux-Powered Robots and Flexible Programming
 

Ok thats like saying that we can use a more powerfull motor because its still a motor. Im just saying allot of teams won't be able to do something like this so you got to weigh how much of an advantage does it give over other teams. however Adam Richards said it best when he said "We don't know. The rules are still subject to change until January 7th, 2006. They might include a laptop with a linux kernel on it next year in the kit and say "Go Forth Students, and program us a LinuxBot!" (as said in IM)" Based on this years rules i think that the rules are pretty clear about just using the RC to control the bot. However by all means go forth and develope this technology im the last one to stand in the way of new advances! Just saying don't make a robot dependent on this and get to the competition just to find its not legal. Get it in writing!!! This is exactly the type of thing you want to check with FIRST on.

Re: Linux-Powered Robots and Flexible Programming
 

I think a better analogy is that if there were a team out there that invented a better motor, they should be able to use it.

You are right, we definetly do need to wait until the new rules are out to be 100% sure.

Re: Linux-Powered Robots and Flexible Programming
 

Guys, we're (or at least Team 245) are not planning to leave anyone behind. The solution costs $200 max (more like mid-100's), and can be FREE if you have junk lying around.

I'm developing all of this Open source, focusing on community-contributed modules and code that anyone can use.

Other teams have used coprocessors before, so I see nothing different here. Analyzing the "Custom Electronics" rules one-by-one, we're in the clear as long as the coprocessor doesn't send out PWMs or communicate to someone in the stands with Wifi in the middle of a match ;)

Re: Linux-Powered Robots and Flexible Programming
 

LOL, in OCCRA (a local, mini-FIRST-like competition), we've had inspectors stop us from putting limit switches on our robot because it's too "dangerous" and carries "live voltage"


But as far as legality, I don't want this to be implemented as a "loophole", it'd be much better if we can get a group of supporters ( you guys with our programming team) to approach the FIRST folks and try to ensure that next year's rules won't deny innovation in FIRST programming.

Re: Linux-Powered Robots and Flexible Programming
 

I think that's a safety thing (i.e. if the coprocessor generates PWM outputs that can override disabled mode, or allows additional robot drivers, etc etc). Technically, it IS the FIRST controller controlling the robot, but just reading output from another device. I've seen teams use PID IC's and other "electronic parts" that have powerful DSPs and powers built-in ;)

Re: Linux-Powered Robots and Flexible Programming
 

My point is that there is nothing new about this other than that it runs linux. Teams have been using auxillary processors in conjunction with the RC as long as i have been involved in FIRST. The rules in previous years have been pretty clear on this. It is a "custom circuit" and as long as it meets the price limits and does not directly output a PWM signal directly to the speed controllers it is legal. They did not mention where the PWM signal comes from. I will assume that this circuit communicates with the RC through the serial port or something like that and Instructs the RC to output the PWM signal. In this case it is legal. If it is intended as a stand-alone replacement for the RC it would not be legal.

Re: Linux-Powered Robots and Flexible Programming
 

So, wait, let me get this straight:

It might actually be legal to go to a used hardware store, pick up a $50 Pentium 1 laptop, load a micro distro & software-of-choice, and then mount it to the 'bot?

That would be wicked awesome! Like, totally! (/me runs out of phrases)

FIRST: Please let us do this! Please! *Gets down on knees and begs.*

Re: Linux-Powered Robots and Flexible Programming
 

Since the human body generates an electric field, shouldn't we be considered too "dangerous," and not be allowed to touch the controls, as we might "damage" the competition RC? :D

Re: Linux-Powered Robots and Flexible Programming
 

Strapping a laptop would be the most cost-effective and featureful way of doing this, if your robot's weight permits (a nicely stripped P2 laptop weighs just slightly over 1lb).


As far as current development, I've began coding the TCP/IP modules framework, along with a true Python serial backend as opposed to this piping picreader hack ;). I expect to have fully functioning betas of everything within about two weeks if school permits :)

Re: Linux-Powered Robots and Flexible Programming
 

Did you implement TCP/IP through the PIC18's serial ports or through the digital I/O?

George

Re: Linux-Powered Robots and Flexible Programming
 

From what I understand, the TCP/IP is for connecting to the laptop that is connected to the RC via the serial port.

Re: Linux-Powered Robots and Flexible Programming
 

have you guys considered handhelds? older ipaqs can probably be had cheaply and are linux capable, then there's the sharp zaurus. i've had an sl5500 for years. now. of course, these lack digital i/o, but if you just want to mount a light weight capable and complete coprocessor i doubt that you could do much better.

Re: Linux-Powered Robots and Flexible Programming
 

TCP/IP (actually, we switched to UNIX named pipes for latency improvements) is for connecting coprocessor modules to the coprocessor serial communicator. Basically, a way so that the community can make prepackaged software modules in the language of their choice, and have it run out-of-the-box for other teams.


We don't use any fancy protocols across the serial line due to severe processing limitations of the PIC18.

Re: Linux-Powered Robots and Flexible Programming
 

Thanks. I was mystified there for that exact reason - the prospect of implementing TCP-IP on a PIC18F is rather ridiculous (to put it lightly). I understand now.

I ask because my team is working on something similar in function but much different in execution - but we're not quite as confident as you are, so we won't be releasing anything until we have a working system. It'll all be GPL'ed (or similar) though as well, have no fear.

Re: Linux-Powered Robots and Flexible Programming
 

Are there any specific things that you (pl) are interested in doing with access to the power of a full computer?

Re: Linux-Powered Robots and Flexible Programming
 

Better image processing is a key factor, but the primary motivation (for us at least) is the challenge and the "cool factor".