1. Botmatch
- Posted by Ralf Nieuwenhuijsen <nieuwen at XS4ALL.NL> Nov 09, 1998
- 594 views
------=_NextPart_000_0038_01BE0B7C.1389C9C0 charset="iso-8859-1" Attached is a package Ive sent to Ck (? sorry, I dont know your real name ?) however, it either never arrived, or he discarded it. Anyways, I did not get any response at that email. So, i'll post it here. Its a working bot-match proggie with an extremely dumb bot. I wont suggest this to be the standard, however, look at it as an (working) alpha. Some interesting techniques. Please, look at the readme.txt And if some one, would please make a bot, that at least, *always* finds the way through the maze.. Ralf ------=_NextPart_000_0038_01BE0B7C.1389C9C0 name="BOTMATCH.zip"
2. Re: Botmatch
- Posted by C & K L <candk at TICNET.COM> Nov 08, 1998
- 547 views
- Last edited Nov 09, 1998
Ralf, I did receive that and responded to it, so maybe the mail is just a tad bit slow...? A few major changes to the web site for anybody needing that info: http://www.ticnet.com/thelesters/eubots Also, I'm about to release the referee so you guys can test your own bots for competition. Later! ck Ralf Nieuwenhuijsen wrote: > > Attached is a package Ive sent to Ck (? sorry, I dont know your real name ?) > however, it either never arrived, or he discarded it. Anyways, I did not get > any response at that email. > > So, i'll post it here. Its a working bot-match proggie with an extremely > dumb bot. > I wont suggest this to be the standard, however, look at it as an (working) > alpha. Some interesting techniques. > > Please, look at the readme.txt > And if some one, would please make a bot, that at least, *always* finds the > way through the maze.. > > Ralf > > ------------------------------------------------------------------------ > Name: BOTMATCH.zip > BOTMATCH.zip Type: Zip Compressed Data (application/x-zip-compressed) > Encoding: base64
3. Re: Botmatch
- Posted by Hawke <mdeland at NWINFO.NET> Nov 08, 1998
- 572 views
Ralf Nieuwenhuijsen wrote: >And if some one, would please make a bot, that >at least, *always* finds the way through the maze.. quick idea on that is you always take lefts (or rights, which ever you prefer) unless you *cant*... this way you will always be hugging a wall, and you will *always* get out... an advanced form of that is leaving a string tied to a stake in the ground, unraveling as you go to let you know if you've been there before... once you have those 2 (somewhat easy) code fragments... piece 'cake :|
4. Re: Botmatch
- Posted by Mike <michael at IGRIN.CO.NZ> Nov 09, 1998
- 527 views
Actually, there a kind of maze where this would not always work. This is one that has a (or part of a)maze structure *not* connected to any exterior walls. If the 'bot attaches to any element in this structure then it will go round 'n round. A solution would be to record your start position and if this position is encountered again then some virtual decision needs to be made to start looking for some wall not previously traversed. Mike. -----Original Message----- From: Hawke <mdeland at NWINFO.NET> To: EUPHORIA at LISTSERV.MUOHIO.EDU <EUPHORIA at LISTSERV.MUOHIO.EDU> Date: Monday, November 09, 1998 3:43 PM Subject: Re: Botmatch >Ralf Nieuwenhuijsen wrote: >>And if some one, would please make a bot, that >>at least, *always* finds the way through the maze.. >quick idea on that is you always take lefts (or rights, >which ever you prefer) unless you *cant*... >this way you will always be hugging a wall, and you >will *always* get out... an advanced form of that >is leaving a string tied to a stake in the ground, >unraveling as you go to let you know if you've been >there before... > >once you have those 2 (somewhat easy) code fragments... >piece 'cake :| >
5. Re: Botmatch
- Posted by Irv Mullins <irv at ELLIJAY.COM> Nov 09, 1998
- 540 views
Cool! My I suggest: when your bot retraces its steps, have it change the color of its tracks so we can see the "dead end paths" more clearly. Irv
6. Re: Botmatch
- Posted by C & K L <candkNOSPAM2ME at TICNET.COM> Nov 09, 1998
- 546 views
Mike wrote: > Actually, there a kind of maze where this would not always work. This is one > that has a (or part of a)maze structure *not* connected to any exterior > walls. If the 'bot attaches to any element in this structure then it will go > round 'n round. A solution would be to record your start position and if Yes, this would happen, so don't make your bot that dumb...
7. Re: Botmatch
- Posted by C & K L <candkNOSPAM2ME at TICNET.COM> Nov 09, 1998
- 535 views
Your bot will have to track where it's been... the screen can't be drawn to, because if another bot encounters your drawn tracks, it could get confused. Irv Mullins wrote: > Cool! > My I suggest: when your bot retraces its steps, have it > change the color of its tracks so we can see the "dead end > paths" more clearly. > > Irv
8. Re: Botmatch
- Posted by Hawke <mdeland at NWINFO.NET> Nov 09, 1998
- 546 views
C & K L wrote: >Your bot will have to track where it's been... >the screen can't be drawn to, because if another >bot encounters your drawn tracks, it could get confused. what about this?: constant BRIGHT = 8, North = 1, South = 2, East = 3, West = 4 function lookdirection(integer dir,sequence pos) --pseudocode folks... integer peekpxl case dir of North: peekpxl=read_pixel(pos+{ 0,-1},1) South: peekpxl=read_pixel(pos+{ 0,+1},1) East : peekpxl=read_pixel(pos+{ 1, 0},1) West : peekpxl=read_pixel(pos+{-1, 0},1) else Error("sumfin wrong") end case if peekpxl > BRIGHT then return peekpxl - BRIGHT else return peekpxl end if end function now, when you set a marker to where you have been, just set it to the color that is on the floor where you are standing + BRIGHT, and your bot will ignore it when you call lookdirection() does that solve the prob??? :)
9. Re: Botmatch
- Posted by Irv Mullins <irv at ELLIJAY.COM> Nov 09, 1998
- 540 views
On Mon, 9 Nov 1998 09:08:04 -0600, C & K L <candkNOSPAM2ME at TICNET.COM> wrote: >Your bot will have to track where it's been... the screen can't be drawn >to, because if another bot encounters your drawn tracks, it could get >confused. > It seems that we have the following situation: background = BLACK maze walls = BLUE finish = RED Now, by my estimation, that leaves at least several colors for the bots. If we run 4 bots at a time, we can use 4 different colors for the bot, plus 4 more for the backtrack, and still have some left over. Using Ralf's bot handler, your mouse will never encounter any other tracks - because the handler only notifies when there is a collision with a wall or the finish line. In addition, if we intend to actually have battling bots, it will be necessary to differentiate between the actual bot position, and the bot trail. No use blasting away at bot vapor trails, is it? Nor bouncing off them, either. Irv
10. Re: Botmatch
- Posted by Ralf Nieuwenhuijsen <nieuwen at XS4ALL.NL> Nov 10, 1998
- 544 views
>Your bot will have to track where it's been... the screen can't be drawn >to, because if another bot encounters your drawn tracks, it could get >confused. Using the screen is slow. Its better to have the maze inside a sequence (maybe on a smaller scale) and display whatever is happing in full glory using sprites and all.. Plus, simply overload pixel (), get_pixel (), poke (), peek () and call () would eliminate all chances at cheating. Ralf
11. Botmatch
- Posted by Lewis Townsend <keroltarr at HOTMAIL.COM> Jul 07, 1999
- 544 views
Ralf, I ran Botmatch and it's really cool but I couln't figure out how to "win". I also read your "todo.txt" file and was very impressed with what you plan to do. I'm looking forward to when I can script something like: "go hit the guy in the red shirt 3 times on the head and push him off the nearest cliff that is over 30 m high and has rocks at the bottom" :P later, Lewis Townsend _______________________________________________________________ Get Free Email and Do More On The Web. Visit http://www.msn.com
12. Re: Botmatch
- Posted by Irv Mullins <irv at ELLIJAY.COM> Jul 09, 1999
- 571 views
On Wed, 07 Jul 1999, Lewis Townsend wrote: > Ralf, > > I ran Botmatch and it's really cool but I couln't figure out how to "win". > I also read your "todo.txt" file and was very impressed with what you plan > to do. I'm looking forward to when I can script something like: > "go hit the guy in the red shirt 3 times on the head > and push him off the nearest cliff that is over 30 m > high and has rocks at the bottom" > :P That command is already built into the operating system. Just put on a red shirt, and type "WIN". Irv
13. Re: Botmatch
- Posted by Ralf Nieuwenhuijsen <nieuwen at XS4ALL.NL> Jul 09, 1999
- 557 views
- Last edited Jul 10, 1999
> I ran Botmatch and it's really cool but I couln't figure out how to "win". I commented it out, so that it shows how the bot works without a end-place and to emphasize the fact that Lucius' bot (he wrote it) doesn't know where the end-point is, however does make a choice. (read: it doesn't cheat) > I also read your "todo.txt" file and was very impressed with what you plan > to do. I'm looking forward to when I can script something like: > "go hit the guy in the red shirt 3 times on the head > and push him off the nearest cliff that is over 30 m > high and has rocks at the bottom" Oops, I still have to remove that file, I don't see me working with my botmatch engine again. If I were to do such an engine, I would write it from scratch though, in some areas its very hard to add some flexibility to the engine without totally changing the design. Therefor, I prefer, I were to work on such an engine (not in the near future) that I would 'reconsider' the basic design of the engine, in the light of 'flexibility' .. actually I would rather see me writing a more 'practical' engine designed for one type of game/program, which is much more feasible in a real-life game and much more 'rewarding' during the procces of programming and therefor I would be more likely to finish such a project. (I admit with shame) Ralf N. http://www.xs4all.nl/~nieuwen