It Is All Your Fault! Hooked on Mazes.

[Burkhardt]

Couple of weeks ago I saw Jack Jarvis' post about cylindrical maze machining files and that got me thinking about an easier and flexible way of creating tool path files for various mazes (actually it cost me some sleepless nights).

Unfortunately, the popular maze generating algorithms I found on the 'net do not lend themselves easily to convert to router bit paths and honestly some of the math was way over my head. However, I figured out my own algorithm that should be easy to convert into g-code or sbp files and allow for flat, cylindrical repeating or 4-sided repeating patterns with selectable outlines (that latter 2 would need an indexer). So far my test code only creates a screen graph but even that looks pretty cool. If somebody wants to play with it, here are the files

If you want to machine a 20-ft size maze and drive the puzzler crazy you could do this.

A more practical design would be that...

Is that worthwhile to develop into a real CAM specialty program? If there is enough interest I may go for it.

[myxpykalix]

G,
HA HA HA....(actually it cost me some sleepless nights) it looks like this is one of those nights.

I had seen a file made by TJ and was interested in info on it and where the supplies like the glass cylinder and sizes and had not heard anything. There have been other maze files but obviously the larger you make it the bigger it has to be.
This is why i was interested in the cylindrical maze files because you could make it bigger (going from 1" to 4" diameter) and it wouldn't make it so unweildy that you couldn't have it sitting on the coffee table or throw it in a bag to take for the kids to play with in the car.

For me what is going to dictate sizes is where you can locate the glass cylinder, end caps and ball bearings (or marbles) for supplies.
Because it seems to me that material size, depth of the trough cut, and size of the bearing along with clearance of the glass cylinder are all things that need to be taken into account when figuring this out.

That was why i was hoping to get info from Shopbot or TJ to see if those files were available or info n where to get the supplies.

I haven't looked at your files yet but if it is something i can read and try to make something on my indexer i may give it a try.

[myxpykalix]

G,
I downloaded the 2nd file you posted, a exe file and upon opening it it shows a representation like in your picture but i see no type of controls or anything that would allow you to change the maze, or save it or do anything? I even tried to manually move in the screen thinking i could do that but nothing...

I'll go try the first file. Not sure what to do

[steve_g]

To be a usable maze generator for non-programmers it would need (in my opinion) the following features...
· The ability to set gross dimensions
· be able to set path width
· select start and stop points... or at least some parameters like start at top, end at bottom.

A neat feature would be the ability to flood fill a given perimeter vector. This would let you select an outline of say a cow, select the mouth as start point and the other end as finish. making it an educational aid for vet schools... Or an outline of Texas and help Davey get to the Alamo!

Since vector creation is so easy with bitmap tracing, you wouldn’t even need that feature built into the program!

These are my quick ideas and I’m sure others will add to it!

SG

[khalid]

The first one is amazing. Great programing skills you have.

[zeykr]

Nice!
I think it would be fun to work with. I agree with Steve about his points especially the fit to perimeter vector.

[scottp55]

I was playing with the idea of a ping-pong paddle size one for children who have problems with fine motor skill problems (saw one on the forum) but don't have the programming skills yet. Yes, It would be very helpful.
Love your ideas, but have enough problems with sleeping now with Desktop "Ideas". Shopbot should give you the 'education" discount so you can play directly with a machine.

[Burkhardt]

G,
I downloaded the 2nd file you posted, a exe file and upon opening it it shows a representation like in your picture but i see no type of controls or anything that would allow you to change the maze, or save it or do anything? I even tried to manually move in the screen thinking i could do that but nothing...

I'll go try the first file. Not sure what to do

Well, this is just an experimental program that has no parameter input (all settings are still hard coded in the source text) and no output other than the image. I just posted it because I think it is cool watching the maze develop. However, I can already program mazes with 90-degree or 45-degree directions, selectable crumminess and selectable branching probability. The algorithm also allows for closed surfaces, that is if an end of the matrix is reached the path pops up at the other end (so you can cut a closed maze into a cylinder or 4-sided stud). I could do the cow or Texas outline but the input of perimeter cells is still manual. I would have to do something to load an image file.

But that is why I was asking about interest. So far this was just a fun problem solving exercise but to make it practical I would have to invest a week or two writing code for a user interface and g-code/sbp output.

[dana_swift]

Looks to me like you are having fun! Its great to see other people who like to combine computer programming with CNC machines. Algorithms are fun to develop, as they require so much pondering.

I have also messed with the maze problem. One option on a program is to allow multiple solutions, another is to allow arbitrary start/end points so the maze cannot be solve using a simple algorithm like the right hand rule. (Always touch a wall with your right hand, it automatically solves all mazes with exterior entry-exit points).

Should you allow islands? Maze paths with no way in/out?

Perhaps only one exit, but many start points. So its not a matter of learning "the" solution.

The branch probability should be programmable, so at any point in the maze the chance of an optional path to consider can be specified.

Lots of fun to think about-

Looking good!

D

[myxpykalix]

I can offer you ideas and suggestions and can try to cut something on my indexer but i have no skills at programming something like this. I leave that to you "Nerds" (an affectionate term) who understand how to do that kind of stuff as it is way over my head.

I like to hear you and Dana talk about algorithyms, ect but don't understand a word of it (kinda) but do have a indexer and if you make something and want it tested i'll do that.