Tuesday, June 2, 2009

Virtual Dog Evolves via Virtual Genes

from (http://www.syntheticthought.com/st/artificial-intelligence/37/154)

Researchers at the Korean Robot Intelligence Technology Lab have developed a virtual dog that lives within a computer simulation. The dog, known as Rity, possesses virtual chromosomes that control different traits of the dog. When a genetic algorithm is applied, the chromosomes randomly mutate so that over many generations different behavioral characteristics appear.

The picture below is a graphically illustrates some of the 1,764 individual genes in a dog, with each gene’s value stored in a floating point type. The darker bands represent higher gene values while red represents lower values for that particular gene.

virtual-dog-genemap

The genes are grouped into chromosomes which are further grouped into three units. The following shows the units along with the contained chromosomes:

The Motivation Unit

  • Curiosity
  • Intimacy
  • Monotony
  • Avoidance
  • Greed
  • Desire to Control

Homeostasis Unit

  • Fatigue
  • Hunger
  • Drowsiness

Emotion Unit

  • Happiness
  • Sadness
  • Anger
  • Fear
  • Neutral
Physorg offers a detailed explanation about the relationship between chromosomes and behavior:
“In Rity, internal states such as motivation, homeostasis and emotion change according to the incoming perception,” Kim said. “If Rity sees its master, its emotion becomes happy and its motivation may be ‘greeting and approaching’ him or her. It means the change of internal states and the activated behavior accordingly is internal and external responses to the incoming stimulus.”
The internal control architecture processes incoming sensor information, calculates each value of internal states as its response, and sends the calculated values to the behavior selection module to generate a proper behavior. Finally, the behavior selection module probabilistically selects a behavior through a voting mechanism, where each reasonable behavior has its own voting value. Unreasonable behaviors are prevented with matrix masks, while a reflexive behavior module, which imitates an animal’s instinct, deals with urgent situations such as running into a wall and enables a more immediate response.
The researchers have programmed the creatures to evolve over time via gene mutation and virtual reproduction. The article explains this mechanism:
While genes are inherited, mutations may also occur. The nature of the genetic coding is such that a single gene can influence multiple behaviors, and also a single behavior can be influenced by multiple genes.

To demonstrate an artificial genome, the researchers used their evolutionary algorithm to generate two contrasting personalities for Rity - agreeable and antagonistic - and compare Rity’s behavior in the different cases. Running the algorithm through 3,000 generations took about 12 hours to generate a genome encoding a desired personality by a Pentium 4, 2 GHz processor. For comparison, the researchers also used manual and random processes to generate genomes with agreeable and antagonistic personalities, though neither outperformed the evolutionary algorithm in terms of personality consistency and similarity to desired personality. Finally, the researchers also verified the accuracy of the evolutionary genome encoding by observing how the artificial creature reacted to a series of stimuli.
Researchers plan on expanding their algorithm in the future to alter genetics based on creature experience, so in effect the user could create virtual dogs sharing similar genetic lines. This is not unlike what is practiced by breeders of real life canines.

This video shows a happy dog interacting with a dog with a poor disposition: