Pinky and the brain

Introducing my two test rats-



which.. then killed the maze window (somehow).

Errors

I ran into some problems with the webots software as I couldn't re-compile the controllers (even without changing any of the sample code). So I sent off an email and waited for support instead. Just incase on the odd chance someone finds my blog, does rats life and also gets this same error, its pasted below:

<< Error would be pasted here but unfortunately I got locked out of my apartment after posting this and couldn't actually get to my laptop as planned. WILL UPDATE SOON >>

The fix:
They got back to me today and said that my java RE was up to date but my SDK (javac/compiler) was old (I had 1.6 as runtime and 1.5 as SDK). Hopefully this will work, I will update later today.

---

Meanwhile whilst waiting for them to get back to me on the error I have been reading their sample code and also partly trying to get my head around this paper:
http://portal.acm.org/citation.cfm?id=1141144

I have an e-link to the actual document but I think it requires login, not sure if the UCSD library's ejournal database has access to it but my ANU one does at least. The paper is called 'Strong planning under partial observability' and is published under Elsevier Science's Artificial Intelligence journal.

In effect it describes mapping in the (realistic) case where you don't actually know your whole environment very well. This definately applies to rats life!
On the downside the paper is a *LOT* to read.

Example of one in real life

Some videos

Point attribution and distance sensors:



A match between Krysicka & Ratatouille 08.06.03

The competition!

The main website for the competition is described here:
http://ratslife.org/

Basically the competition takes place both in a simulated virtual environment as well as a real one (though the real case is too slow so for actual rankings they just use the simulated). Real case uses lego.

Robots are placed in to a maze (random) in pairs- whoever survives longest wins. Robots have a simulated 'power meter' which they must replenish at power sources scattered around the maze. Once you replenish the source deactivates for a set amount of time before it can be used again (so you can go look for a new one in the mean time).

Each robot is equiped with
-8 Distance Sensors
-10 LED lights (just for fun?)
-1 Camera
-2 Wheels + Accelerometer
Each

Sounds like fun.