Anna Pawlicka

Programmer. Hiker. Cook. Always looking for interesting problems to solve.


About | Archive | Talks

Bazinga the Retarded Dog

01 Feb 2013 | AI, Aibo, artificial curiosity, reinforcement learning, robotics, Webots

Since October last year I have been working on a robot simulation (Sony’s Aibo) trying to implement artificial curiosity and learning algorithms. Why? Final-year project. But also a very interesting and challenging problem to solve. I am working under supervision of Dr Chrisantha Fernando, at Queen Mary University of London.

As of now, Bazinga is not the smartest dog out there:

I am now looking into Q-Learning temporal-difference control algorithm to make use of the red ball. The idea is for the Bazinga to be rewarded if it gets closer to the ball. One problem I am currently trying to solve:

The data structure to hold state-action values, e.g. 3-dimensional array where first and second indices are x and y coordinates and third index is a possible action from current state. Considering that Bazinga’s movement is controlled by 12 joints, this would be quite large array. I could simplify it by replacing joint movement with simple North, South, West, East directions. This seems like oversimplifying though.

Time to read more about reinforcement learning.

P.S. Yes, I am a big fan of the Big Bang Theory.


Older · View Archive (25)

If you haven’t installed IntelliJ yet, you should do it now.

Everyone has their favourite editor/IDE – mine’s IntelliJ. Why?

Newer

A not very successful attempt at red ball fitness function.

I decided to try a very simple genetic algorithm, with my own methods to copy and mutate individuals, and a very basic fitness function that, in theory, should try to minimize the distance between Bazinga and the red ball.