h a l f b a k e r yThese statements have not been evaluated by the Food and Drug Administration.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
This would be a program that sword-fights without the traditional sequence of pre-sequences. Instead it can move all of its joints to any other position; it can see what the other AI is doing.
The AI uses a NegaMax tree, with the evaluation function changed as the result of a Finite State Machine, E.G.
favoring Attack or defense or retreat. They get rated apon how many AIs they defeat and their ratings
Through all the possible moves, within 25 frames e.g. saying it has 6 joints which can go 360 different positions, it can go 6 * 360 * 2 * 25 = 108000 possible moves per second.
There are 180000 possible moves for 1 second, it checks one tree of learned moves (usually faster, and it allows multi-second moves) or it can use the second tree which once it finds a particularly good move it adds it to the first tree, and records it for later. To stop especially long games, the tree may have to search further away from the target when the time exceeds one minute then stop once it reaches ten it would require to search 64800000 posible moves over ten minutes. It also records counter moves and records them, so next time it can counter the move if it sees it agAIn, it can also create counter-countermoves agAIn using the trees
You can also play agAInst this AI, allowing you to teach it various moves and counter moves, it can also run MASSIVE trees using up any spare clock cycles. A Players health goes down by how many moves manage to hit, i.e. it had no counter moves for.
Moreover, if you get bored you can see the trees and the generated swordplay, playing 24-7 while at least two computers were up. It would require minimal traffic for each game. (it would only require two numbers one from 0 to 6, one from 0 to 360.)
(i'm new here- it's my first idea that got any reply other then from the "autoboner")
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Annotation:
|
|
Right, AI, not AL. I keep messing that up. We need a font where the capital "I" has those bars on both ends. |
|
|
AI Swordfighting used to be the serif in this old town. |
|
|
{overpanic} Hopefully :-) |
|
|
Sp: Swordfish - but I didn't bone it! + |
|
|
True sword-fighting is hard to do with the current interfaces... though if it had a peripheral for it (like the one ms is coming out with - essentially an upgraded wii-mote) it would work well |
|
|
It is not clear to me how you would "play". Would the chief attraction for a player be trying to beat "Swordfighting Al"? Or would the attraction be trying to write the learning algorithms, akin to Robot Battle. Or would one sit back and watch beautiful swordfighting artificial intelligences as a sort of screensaver / meditative aid? |
|
|
It's not in "games" it's in AI, so seems to me like it's a learning project... the AI can optimize sword fighting techniques or something along those lines. Into a game... I'd like to be able to sword fight a computer - great training if it's realistic (although you'd need some sort of feedback for blocks, etc.) |
|
|
I suspect that as with any created environment which is learned by the AI, one will learn more about the created environment than anything real. Which is why the process of learning seems to me the most interesting. |
|
|
Trees fighting sounds like a good idea. It's about time they were able to defend themselves against chainsaws, ivy, and ash borers. |
|
| |