Slashdot Mirror
Linux Powered Robots
Al writes "We all remember RoboCup2000. LinuxWorld.com.au has an article on how Linux was used by different teams to optimise communications, hackup wavelan ethernet and build in sensory and visual intelligence to the Robots. Is it too early to ask for christmas presents?" Anyone else digging on Comedy Central's Battle Bots? I wish they'd ditch the lame sports comentary and replace it with purely technical stuff. And then watch robots smash each other. Somehow I doubt that you could build a winning battle bot based on Linux: it just strikes me that things like hard drives don't respond well to axes and chain saw blades.
Now the right :) way to do it would be to give each part of the bot a simpler controller (ROM-register, or even analog design). For example, the chainsaw system would be independent of the propulsion system. Advantages would be:
- No central 'brain' to take out. Chainsaw would still be useful if the bot was sitting in one place because the propulsion got smashed.
- Simple controllers would be robust and allow redundancy.
- Greatly reduced power requirements. This means power could be redirected to lugging around heavier arms and armor.
- Cheaper. This means money could be redirected to bigger and better arms and armor.
- This may allow the use of the coolest anti-bot weapon of them all: EMP. Crack the armor, pulse, and your opponent is disabled, or at least reset. Since your simple analog designs are basically EMP-proof, and simpler digital designs are easier to shield if they need it, you come out on top.
The only disadvantage I can see is:- Need to know how to design closer to the hardware. We're talking about circuit design and control theory as opposed to C programming. EE background as opposed to CS background.
Come to think of it, this would be a cool way to design a real robot, too.It was actually quite encouraging to see so much acceptance of Linux. I really didn't expect it -- I asked a few people, just out of curiosity, what operating system they were running, and the more people I asked, the more positive responses I received to Linux. Anyone that really knew anything about Linux was using it. (maybe that's why they knew something about it ... ??? duh ...) But the real kicker was the overall attitude people had towards it. They just loved it. No one had techno-rage at a Linux machine. :)
As a member of Cornell's Robocup F180 team (1999 and 2000 champions) I can say that Robocup and Battle Bots are completely different. As I understand it, Battle Bots are human controlled. I know the robots in RobotWars are. Robocup's robots are COMPLETELY autonomous. Once a game starts, team members have no control over the robots.
In my team, we've talked/joked about entering Battle Bots, and I'm sure it would be exciting, but there is no way we would be willing to risk our robots to those saws, etc. Our robots are custom machined and incredibly expensive. (We estimate $5k a piece) There is no way we can throw that at a guy with a joystick controlled buzz-saw...
That being said, Robocup is only getting more exciting, though I agree, we could usually do without the commentary.
Exiting house...done.
Moving down stairs...done. Raising arm to obtain mail... Segmentation Fault. Core Dumped."Ancillary does not mean you get to rule the world." --U.S. Circuit Judge Harry Edwards, speaking to the FCC's lawyer
There are several places where computer controls would be quite useful, all the same.
For instance, if there were sensors to indicate when that slashing blade should go down so that it would only happen when the Bot "saw" another Bot in range, that would make the attacks "more accurate."
As it stands, a whole lot of the contest comes out of how successfully the human can percieve the precise orientations and relative locations of the "Bots." If your depth perception is crummy, then the battle will go badly.
There are other mechanisms to get to the same goal; having cameras at critical locations so that the human can see precisely what you're pointed at would "do the trick."
But the more complex the set of things that the Bot can do, the more useful it can be to automate control over some of those things.
Obviously it won't involve a vulnerable IDE disk drive; I'd expect such a system to use an embedded controller using rather rugged hardware. As it stands right now, if servo cabling gets cut, a Bot will be crippled, which is essentially no different.
I agree that part of me would be more impressed if the "Bots" were truly autonomous; it is not at all obvious that that would result in entertaining TV, unless we could go a few generations further to "Robots" where the designers were essentially devoted to programming entertaining strategies.
If you're not part of the solution, you're part of the precipitate.
If someone built a Battle Bot that ran on Mac OSX
and that bot became champion would that make
Mac OS X the most powerful OS?
http://Lenny.com
What if the Linux battlebot was running a program that attempts to break into it's opponent electronically and defeat it by hacking it's O/S rather than smashing it with a chainsaw?
The difference between Canada and the USA is that in Canada healthcare is a right and gun ownership is a privilege.
I don't understand Monsieur Le Taco's comment. Linux wouldn't make a winner because it runs on a hard-drive? Um, er, linux is software isn't it? I imagane that it could be loaded onto something a bit sturdier...
...Survival Research Labs!
---
Zardoz has spoken!
Oper on the Nightstar
Who says you need hard drives? Just use Flash RAM to hold the system. There are ways around most normal system considerations, but they're usually expensive. Of course, this isn't a hobby for cheapskates. Just how much do those Lexan exteriors that some of the bots on Battlebots use cost?
If it's for-profit but free, you're not the customer -- you're the product (e.g., the Slashdot Beta's "audience").
How about the minds here at Slashdot getting together to actually design a robot to go into the ring on Comedy Central?
Here is my concept: A bot that is controlled by both remote control AND an onboard computer system. The onboard system would have a connection to the controller's laptop via wireless networking. Place a "quickcam" on the bot so the controller gets video feedback from the ring (one of the big problems that I have seen on the show is when the robot is turned around opposite to the person controlling it, the person in control has difficulty handling the reversal in the controls.
The answer is to give the person in control a first person perspective.
Then setup the keybindings on the laptop to be the same as those used in Quake. And give the bot a chainsaw.
You would then have the ultimate bot/human interface because the human would be adapted to the controls through years of training for deathmatches in Quake.
Brought to you by Frobozz Magic Penguin Fodder.
Rodney Brooks has been working on his Cog project for the last several years, but before that he worked on a very similar idea to yours, called the subsumption architecture. A good quick overview can be gotten from one of Brooks's early papers, Elephants Don't Play Chess.
WWJD for a Klondike Bar?
is that they are not robots. They're just remote controlled machines. It's still a great show, but I'd be more impressed if they had a mind of their own and not controlled by a human.
Misfit
The Slashbot has a very advanced artificial intelligence, and it has even come close to passing the Turing test on several occasions. When humans are presented with text output from the Slashbot, it initially appears as if the text was produced by an intelligent lifeform. However on closer inspection, it can be seen that the Slashbot achieves it's pseudo-intelligence by recycling a series of tired old arguments such as "information wants to be free" and "many eyes make bugs shallow". Detailed analysis of the Slashbot's literary compositions show that the Slashbot is nothing but an over-opinionated, insecure, self-righteous bore.
My Slashbot can also interpret and respond to text-based information which is supplied to it. Unfortunately, I have been unable to train the Slashbot to stop responding to blatant "troll" input. If I feed the Slashbot data of a form such as "Security is only possible through security", my Slashbot is unable to resist outputting a tedious monolgue detailing the flaws in my argument. I have been unable to stop the Slashbot responding to troll input, even by applying "YHBT YHL HAND" input during the neural network training. A Sourceforge page for the Open Source Slashbot project can be found here.
Thank you.
If you ran Windows on your box, it wouldn't even need to defeat the other bots. It would just absorb them, and become stronger.
Of course, that strategy would only work for a little while. Eventually your robot would become so bloated that it could easily be defeated by a smaller, nimbler bot.
OtakuBooty.com: Smart, funny, sexy nerds.