[FRIAM] solving mazes

Tue Mar 2 14:00:59 EST 2021

I want to push this conversation about mazes a bit.  So, we a visual creatures and we tend to understand mazes as a visual problem, not a tactile or olfactory problem.  The rat brings to a maze all sorts or skills that humans do not.  So wall following is just a matter of putting the vibrissae on one side of your face on a wall and keeping them there until you get somewhere.  You have all pointed out the flaws of wall following  in the case that there is a maze hidden within a maze, in which case the rat will never enter the surrounded maze.  

However the complications of introducing olfaction in to our understanding of the rat’s behavior are myriad.  Let’s put a side the possibioity that the odor of the food diffuses down the maze.  Let’s put aside the thought that stigmergy is at work here because the rat-runners are not expunging all odor cues from the maze between different rats, or different runnings of the same rat.  Let’s ONLY consider the cues that a running rat leaves for himself.  Olfactory cues being what they are, I assume that a running rat ALWAYS knows if he has been to a particular point in the maze, and how recently he has been there.  He also knows what direction he was running whenever he encounters his own trail.  And the rat always knows whether he has found food or not.If you add those skills to your AI maze running kit, don’t you add a lot? 

So, for instance, add to wall following some rules about what you do if you encounter your own odor going in the opposite direction without a trace of food odor.  That might lead you to change walls, in which case you would FIND the maze within the maze and eventually the maze-within-the-maze-within-the-maze .

Do any of you remember piglet tracking hefalumps?  So add to your conception of the maze that you are solving it in the snow and that you leave footprints showing your direction of walking and little bits of crème brule falling off your beard where ever you go after you have been fed.  

Eh?

N

Nick Thompson

 <mailto:ThompNickSon2 at gmail.com> ThompNickSon2 at gmail.com

 <https://wordpress.clarku.edu/nthompson/> https://wordpress.clarku.edu/nthompson/

From: thompnickson2 at gmail.com <thompnickson2 at gmail.com> 
Sent: Sunday, February 28, 2021 11:54 AM
To: 'The Friday Morning Applied Complexity Coffee Group' <friam at redfish.com>
Subject: RE: [FRIAM] solving mazes

Hi Roger, 

Here is a picture of a typical rat maze.  I was going to crop it but I thought the image of a former English major trying to manipulate technology was too funny not to send along. Notice that it’s all t’s. 

You wrote:

If the graph of the maze contains loops, then wall following will get trapped in the loops, in fact, I think each loop defines a wall following domain, and to solve the maze you'll have to detect when you've completed a loop and switch to following the opposite wall when it's one you haven't already followed.  If you just switch to the opposite wall at random, you'll most probably end up on the same loop going in the opposite direction.

This is the kind of thing that rats should be extremely good at.  If there is one thing a rat should know it’s where it’s been and how recently and even, how often.  

Nick 

Nick Thompson

 <mailto:ThompNickSon2 at gmail.com> ThompNickSon2 at gmail.com

 <https://wordpress.clarku.edu/nthompson/> https://wordpress.clarku.edu/nthompson/

From: Friam <friam-bounces at redfish.com <mailto:friam-bounces at redfish.com> > On Behalf Of Roger Critchlow
Sent: Sunday, February 28, 2021 11:03 AM
To: The Friday Morning Applied Complexity Coffee Group <friam at redfish.com <mailto:friam at redfish.com> >
Subject: Re: [FRIAM] solving mazes

Okay Nick, 

So you draw the maze as a graph where the nodes of the graph are positions in the maze that can be occupied, and the edges of the graph are one step open paths between adjacent positions.  Easy to draw on graph paper, here is a site which generates graph paper pdfs for printing: https://incompetech.com/graphpaper/, just fill in the cells that are walls and draw lines to connect the remaining open cells.  Each open cell is a node, each line connecting adjacent cells is an edge.

The graph of the maze must be connected or parts of the maze will be cut off from each other and no algorithm will ever work if it starts in the wrong connected component.  Nasty trick to play on a rat.  

If the graph of the maze contains loops, then wall following will get trapped in the loops, in fact, I think each loop defines a wall following domain, and to solve the maze you'll have to detect when you've completed a loop and switch to following the opposite wall when it's one you haven't already followed.  If you just switch to the opposite wall at random, you'll most probably end up on the same loop going in the opposite direction.

Just to demonstrate how graph properties end up trapping the unwary, an open courtyard in your maze ends up being an area of fully connected graph, which is all loops all the time, which is a counter example to my proposal that "each loop defines a wall following domain" since lots of the loops in the courtyard will have no adjacent walls at all.  So I must have meant something subtler than loop, something that is a loop that has walls on both sides.

-- rec --

On Sun, Feb 28, 2021 at 9:39 AM Jochen Fromm <jofr at cas-group.net <mailto:jofr at cas-group.net> > wrote:

Jamis Buck is very interested in mazes. He even wrote a book about it named "Mazes for programmers"

https://www.jamisbuck.org/mazes/

It contains algorithms for generating and solving mazes

https://pragprog.com/titles/jbmaze/mazes-for-programmers/

-J.

-------- Original message --------

From: cody dooderson <d00d3rs0n at gmail.com <mailto:d00d3rs0n at gmail.com> > 

Date: 2/28/21 01:10 (GMT+01:00) 

To: The Friday Morning Applied Complexity Coffee Group <friam at redfish.com <mailto:friam at redfish.com> > 

Subject: Re: [FRIAM] solving mazes 

I am assuming this is a 2D maze. Wikipedia does a better job at explaining the problems with wall following than I can. 

If the maze is not simply-connected (i.e. if the start or endpoints are in the center of the structure surrounded by passage loops, or the pathways cross over and under each other and such parts of the solution path are surrounded by passage loops), this method will not reach the goal.

On Sat, Feb 27, 2021, 1:48 PM <thompnickson2 at gmail.com <mailto:thompnickson2 at gmail.com> > wrote:

Hi, All,

Due to a review I have been working on, I have been dragged back into thinking about maze learning in rats.  Any animal I have ever known, when confined, will explore the boundaries of its enclosure.  Cows, for instance will beat a path just inside the barbed wire that encloses them.  So a maze is not only a series of pathways but it is also an enclosure.  If the rat puts his left whisker against the left wall of the maze, he will eventually get to the goal box, right.  It works with the Hampton Court Maze.  On the second run, he can now use odor cues, such that any time he encounters his own odor both entering and leaving a passage way, he should just skip that passage way. 

So I am wondering, you topologists (??) out there, how general is the statement, “every maze is an enclosure”  and what is the limitation on the idea that any maze can be solved by putting your right or left hand on a wall and continuing to walk until you find the goal or are let out of the maze.  Now I should quickly say that rat mazes are usually composed of a series of bifurcating choice points, where the rat can go either left or right. Some choices lead ultimately to dead ends.  In sum, a runway in such a maze can go straight, turn R or L without choice or form a T with a right or left choice.  My intuition is that no such maze can be designed that does not permit the boundary following strategy. 

Nick 

Nick Thompson

ThompNickSon2 at gmail.com <mailto:ThompNickSon2 at gmail.com> 

https://wordpress.clarku.edu/nthompson/

- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
FRIAM Applied Complexity Group listserv
Zoom Fridays 9:30a-12p Mtn GMT-6  bit.ly/virtualfriam <http://bit.ly/virtualfriam> 
un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/
archives: http://friam.471366.n2.nabble.com/

- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
FRIAM Applied Complexity Group listserv
Zoom Fridays 9:30a-12p Mtn GMT-6  bit.ly/virtualfriam <http://bit.ly/virtualfriam> 
un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/
archives: http://friam.471366.n2.nabble.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20210302/75abe885/attachment.html>