Re: Robot challenge -Reply -Reply
Irv-
>>> Irv Mullins <irv at ELLIJAY.COM> 11/05/98 01:32pm >>>
On Thu, 5 Nov 1998 11:39:12 -0600, Jay Turley <jturley at MDC.COM>
wrote:
>Imminent collision with the wall - minimum. Don't trash the wall
>when colliding. Also helpful if they could detect collision
>with another 'bot (for later bot wars)
You need to specify a color for bots, then, so that other bots can sense
them. Walls are yellow, start is green, end is red, maybe bots are white?
Then we can dispense with any fancy movement-tracking functions (like
scanning the sensor area for pixels changing to the background color
again.) Or maybe this is what you want?
>Again, up to the coder. Do you go for simple and fast, (head banger)
>or complex and slow (chess master) mode?
Should we provide a hook for polling then? I.e. a single point of
interface into our bots that allows us to perform one action per clock
tick? There would have to be some time limits on a bots move (simulation
of real-time), or else you could design a bot that could just sit there,
analyze the entire map, and then plot a course to the finish, while
another bot just looked ahead, saw a wall, then turned left.
Maybe:
Clock Ticks Action
------------- --------
1 sense up to ten pixels away in one direction
2 sense 11-20 pixels away in one direction
1 move up to 5 pixels
1 turn 45 degrees either direction
or something like that, anyway, so that our bots have to respond to a
clock_tick function by checking our internal state, and then making a
choice of what to do, and either returning an action or a "busy" signal if
taking more than one clock tick to perform an action
>Oh. I forgot to add, we need to see the bot moving on screen -
>maybe a nice circle would be enough - some color other than
>yellow/red/green. Get fancy if you wish. I realize that if
>our algorithm is good, the bot will move so fast it will
>be hard to see. Might be interesting, tho, to stick in a
>time delay to see how it works.
>Naturally we'll include a timer at the call and at the exit
>of the 'bot routine, and time them all on the same pc.
Then my info above isn't necesary, because they won't all be moving at
once. Only if there are multiple bots is it necessary to implement a piolling
procedure. I guess I am confused as to whether you want the best
simulation of a bot running through a maze, or the best maze-solving
algorithm implemented as a bot.
>
>Regards,
>
>Irv
- Regards back,
-jay
|
Not Categorized, Please Help
|
|
