More complex maze generation by circular implementation of the grid

It would be great if a circular maze, which is usually more complex to be comprehended by the human mind, could be implemented using the algorithm that is already here. What I have in mind is the following cell structure:

The design above is not geometrically optimal but could be implemented the following way:

• The rows can be represented as a set of concentric circles and columns seperated by lines between two circles.
• The key difference would be that columns will be looped i.e., the last column will be connected to the first column
• In order to maintain an even distribution in the size of the cell, I recommend doubling the number of columns for every row without which there would be an exponential increase in cell size with increase in number of rows. This would greatly add to the complexity and at the same time maintain the cell size. -That would imply, only the rows (number of circles) will be given as input and the number of columns would be : pow(2,rows+2) -The condition for a cell to be start/exit would be row = num_rows-1

I have implemented the above condition but I am yet to commit the changes as visualizing this in matplotlib is not as simple as doing that for a grid. The following is a manual visualization from the raw outputs of generate_maze.

Please do let me know if you’re more interested in this and have ideas about visualzing this implementation in matplotlib

[jostbr]

Okay, yeah, I see what you mean. Might need a little restructure nefore it should be merged withmaster. Reverted the pull request now.

[ThomasThelen]

See my comments on the commit; with the way things are currently done, adding new features is going to be difficult. Can we un do the merge? We should either create a new branch under dev and put this there, or once some changes are made do a pull request to the dev branch