JavaScript disabled. Please enable JavaScript to use My News, My Clippings, My Comments and user settings.

If you have trouble accessing our login form below, you can go to our login page.

If you have trouble accessing our login form below, you can go to our login page.

Artificial intelligence taking on the green pigs

Date

John-Paul Moloney

Zoom in on this story. Explore all there is to know.

First Angry Birds ... tomorrow the world.

Angry Birds Space... just one of the many Angry Birds games now available to the public.

Angry Birds Space... just one of the many Angry Birds games now available to the public.

Sixteen years ago Garry Kasparov vs Deep Blue in computer chess was the ultimate test of man vs machine.

Now in the age of the iphone, it's man vs computer vs bird...or more correctly malevolent green pigs.

This week in Sydney a group of researchers from the Australian National University will conduct what they hope will be the first of many artificial intelligence challenges involving the incredibly popular game Angry Birds.

In what is bound to be one of the most popular elements of the 25th Australasian conference on artificial intelligence, computer programmers will compete on Wednesday in solving 10 new levels of the game.

The challenge for the progammers is to create a system that will solve the levels.

Their results will then be compared to human players, who will tackle the levels on Thursday.

Associate professor Jochen Renz, one of four ANU researchers involved, thinks this first year will be a human's only chance to beat the computer, because the computer programmers have only had four weeks to prepare.

"That's my guess, this will be the humans' only chance. Some of our computer programmers have done a really impressive job already. At the moment we're not looking at the hardest levels, when we get to them, humans have a massive problem," Professor Renz said.

According to Professor Renz, the reason Angry Birds is so addictive - and so appealing as an AI challenge - is more than just the simple fun of catapulting a variety of attack birds at egg-stealing green pigs.

He said it tested a range of sub-disciplines of AI, including computer vision, machine learning, knowledge representation and planning.

"It is about particular problems that exist in the game. The special relationship between the blocks, which ones need to hit another one.

"They [behave consistently], so if you hit a block in the exact spot the same thing will happen again, but you can't test all of these in advance."

Professor Renz said the Angry Birds challenge was more than just a contemporary version of the man vs machine chess battle, it was quite different.

In chess, each move might have 10 or more options, but the challenge was to plan many moves ahead. In Angry Birds there are only a small number of birds, but each shot "could have a million possibilities".

And, strange as it may sound, he said the battle to the death between pigs and birds was closer to the real world than chess and therefore offered new possibilities for developing intelligent machines.

"Chess is really an artificial problem. Who has to solve a situation similar to chess in daily life? But Angry Birds birds is more relevant to real life, if you knock things they will move and fall."

In a surprising admission, Professor Renz, whose fields of expertise are knowledge representation and reasoning, says the green pigs are safe whenever he takes command of the catapult.

"I'm actually very bad at the game, usually the computer beats me."