I would like to model realistic random rodent paths. Obviously I don't just want a random walk, its too chaotic to be useful.
Has anyone done something like this that can help get me started?

4/19/2012 3:36:16 PM

interesting problem. i don't have any insight, but please keep us posted with your results

4/19/2012 3:45:10 PM

Model meaning what exactly? Are you talking about simulating the pathing of a rodent through a maze?

4/19/2012 4:10:46 PM

I could help you out with some information if you provide a little more detail of the problem. What specifically are you trying to model?

4/19/2012 4:16:34 PM

I want to model a rats movement on a plane foraging for randomly dropped food pellets.


The final intent is to model grid cell firing in the rats entorhinal cortex - but first I just want a realistic representation of rat movement to use as input for the grid cells.


I guess my question is are there simple, well known algorithms for modeling realistic movement on a 2d plane? (Obviously there is something out there, its done all the time in video games).

i only* care about position, direction and velocity

[Edited on April 19, 2012 at 4:26 PM. Reason : .]

[Edited on April 19, 2012 at 4:27 PM. Reason : .]

4/19/2012 4:24:04 PM

If you think they are moving purely randomly (i.e. they don't sense the have food until they've landed on a grid cell containing food) it sounds like you want some sort of autoregressive process. The most popular is probably the AR(1) model, because it's the easiest to use and understand.

http://en.wikipedia.org/wiki/Autoregressive_model

4/19/2012 4:28:06 PM

^ Thanks! yeah, the food is irrelevant, in practice it is just to motivate the rats to move but has nothing to do with what I want to model in the end, I just need realistic motion.

4/19/2012 4:31:48 PM

Sounds like that is exactly what you want then. Glad I could help.

There is an entire area of research for this type of phenomenom which are usually referred to in the literature as stochastic processes. They cover just about every type of random process you could imagine. If you need some more suggestions feel free to ask, but I think AR(1) sounds like it will be good for your purposes.

4/19/2012 4:37:44 PM