Project Vocabulary Defined

Patch — A bin of space that stores information about itself and can reference the patches it is adjacent to in the coordinate space. This idea is based upon the work of Michael Resnick and the MIT Epistemology & Learning Group’s STARLOGO project.

Environment — A collection of connected patches.

Individual or "TURTLE" - An autonomous creature. A turtle contains both a local coordinate system and the methods appropriate to reference the Cartesian coordinate system. The turtle moves in three-dimensional space, and projects its location in a two-dimensional representation. The turtle concept is based in part on Ableson & Decessa Turtle Geometry from MIT.

Decision Making Turtle (DMT) — A turtle whose actions in the environment are based on the conditions it perceives and a corresponding lookup into an action table. This lookup procedure is defined through its genome.

Genome — A list of values unique to a DMT that represents which actions are to be taken in each of the conditions defined by the condition list.

Population — A collection of DMTs.

Fitness — A quantitative measure of a DMT’s success relative to the goals of the simulation. One fundamental goal of the simulation was survival of the DMT, with the possibility of additional primary or secondary goals.

Mutation — Random alteration of a genome.

Crossover - Process by which two parents’ genomes are used to determine the genome of offspring. Crossover preserves genetic material of the parents involved.

Reproduction — A process that selects DMTs from the population based on fitness, applies a genetic algorithm using crossover and/or mutation and then introduces the new DMTs into the environment.

Property — An instance variable that allows information to be stored within a patch or a DMT.

Condition - The perceived state of the DMT in combination with the patch it occupies and possibly its neighbors. The condition must be a member of a finite set of possible conditions, the condition list.

Condition List — The finite set of conditions derived by considering a number of properties. These properties can be external to the DMT (i.e. the amount of "food" present in the near environment), and/or internal (i.e. the amount of "energy" an individual has left).

Action - Something a DMT can do. An action is a member of a finite set of actions able to be performed by a DMT as specified within the action list.

Action List — The finite set of actions available to DMTs. In our simulations, typical actions include changing heading, moving, eating, etc.

Goal - The objective as defined by the architect of the simulation. DMTs are not inherently coded with the objective in mind, rather the reproduction of the DMTs is tailored according to a fitness scale based on a goal (or goals) to encourage the emergence of solutions.

 

Main Description Demos Decision Making Evolution Downloads Other Resources Contact

© 2000 Dan Kunkle, Rochester Institute of Technology