Scratch

so that you don't need to download, here is the script for sensing the walls

when I receive [sense v]
delete (all v) of [distances v]
set [drawing v] to [1]
set [distance v] to [0]
set [scanning remembering thingamajig v] to [48]
repeat (24) //because I have 4 sensors
set [distance v] to [60]
go to [movement v] //the movement sprite
point in direction <<[direction v] of [movement v]> + (scanning remembering thingamajig )>
repeat until <<touching [wall v]?> or <(distance)=[0]>>
move [1] steps
change [distance v] by [-1]
end
add (distance) to [distances v]
change [scanning remembering thingamajig v] by [-1]
end

and here is the script for drawing the walls

when I receive [draw v]
go to x:[1] y:[180] //since each drawer takes half the screen
set pen size to [5]
set pen color to [#0000FF]
clear
pen up
set [help counting v] to [24] //since each distances list is 24 items long
repeat (24)
set pen shade to <<item (help counting) of [distances v]>*<[1]+<[2]/[3]>>>
set y to <<item (help counting) of [distances v]>*[4]>
pen down
set y to <<<item (help counting) of [distances v]>*[4]>*[-1]>
pen up
change [help counting v] by [-1]
set pen shade to [0]
change x by [5] //because the pen size is 5
end 

Each drawer does that script twice to match up with the 4 lists of distances. I'm trying to figure out how to fix that when it senses, it draws it in a curved room when it's supposed to be straight. Can someone please help me?

Last edited by TorbyFork234 (2012-04-13 16:49:17)

AtomicBawm3 wrote:

The reason is because you're using purely distance.  If you use purely distance, even if points are on the same line that's perpendicular to your view, they're different distances.  This creates the Fisheye effect.  In order to correct this, take the distance you get when scanning and multiply by the cosine of the offset of the angle (i.e. if you're drawing the part of the wall right in the middle, the offset is 0 degrees and so the distance will simply be multiplied by 1).

How do I find the offset of the angle? I tried turning 1 degree, moving 1 step, and seeing if it's still touching the wall, and if not, take that direction and continue going. but that didn't work.

so the new script should be

add <(distance)*<[cos v] of (scanning remembering thingamajig)>> to [distances v]

?

Also, what's FOV?

Last edited by TorbyFork234 (2012-04-13 17:21:08)

AtomicBawm3 wrote:

Yep, looks like it from the way you've got your scripts set up.    I hope this helps...what you could do is change you FOV from 96 to something more like 60 as well...that will reduce the effect as well.  That would mean at the beginning:
Sensor 1 variable: 30
Sensor 2 variable: 15
Sensor 3 variable: 0
Sensor 4 variable: -15

But that would also reduce the resolution so I won't do it.

bump

no, I tried that and everything up close was fine, but everything far away was all wacky. and then I tried

<<[60]-<[cos v] of (variable)>>*<[60]-(distance)>>

and that didn't work either.

Didn't work. Help anyone?