Super ants
Download
1 / 14

Super Ants!! - PowerPoint PPT Presentation


  • 103 Views
  • Uploaded on

Super Ants!!. Matt deWet & David Robson. Symbiotic Coevolution. Primary research question: “Can heterogeneous teams of evolving agents, who depend upon each other for survival, learn to work together?”. How to test that. Environment needed: Two specialized teams of agents, run by NEAT

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about ' Super Ants!!' - colum


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
Super ants

Super Ants!!

Matt deWet & David Robson


Symbiotic coevolution
Symbiotic Coevolution

  • Primary research question:

    • “Can heterogeneous teams of evolving agents, who depend upon each other for survival, learn to work together?”


How to test that
How to test that

  • Environment needed:

    • Two specialized teams of agents, run by NEAT

      • Different abilities, different roles

      • Can only survive by working together


Our environment
Our Environment

  • Ants!

    • Soldiers & Workers

  • Environmental Threats

    • Spiders

      • These love the taste of worker flesh

      • Controlled by a static algorithm

    • Starvation

      • Great at killing spiders

      • Not so great at gathering food


Our environment cont d
Our Environment (cont’d)

  • The world

    • Bounded grid of variable size

    • Randomly placed food

    • Randomly spawned enemies

  • Movement

    • All entities move at most one space at a time on the grid

    • Movements all take place simultaneously, so no unit has an advantage


The plan
The Plan

  • Sensors

    • Soldiers can see nearby enemies and workers

    • Workers can see nearby food, enemies, and soldiers

  • Desired behavior

    • Soldiers learn to keep foraging workers safe

    • How can we tell?

      • Overall fitness?

      • Inspection


The experiment
The Experiment

  • Control

    • Evolve the two groups separately, then stick them together and see how they do

  • Experiment

    • Evolve the two populations together, observe behavior

    • Variations:

      • Pre-evolved or un-evolved brains.



Current work1
Current Work

  • Current fitness functions

    • Soldiers

      • fitness: Spiders killed

    • Workers

      • fitness: How much food is eaten

  • Some videos!


Multi tiered networks
Multi-tiered Networks

  • Neural network acts as a switch between behaviors

    • Behaviors implemented as neural networks or algorithms

  • Simplifies each network

    • Minimizes inputs

    • Splits large tasks into learnable chunks


Multi tiered networks cont d
Multi-tiered Networks (cont’d)

  • Advantages

    • Intuitive

    • Smaller and less complex networks

    • Generally faster than traditional AI algorithms

  • Disadvantages

    • More human labor-intensive for development/design

    • Some tasks may not be easily divisible


Future work
Future work

  • Shared fitness

    • Reward for colony doing well

      • More important for soldiers

    • Problem:

      • Any shared fitness among all agents in one population is nullified, because only relative fitness is used to determine who reproduces. 


Future work1
Future work

  • Alternate fitness functions

    • Slightly more engineered

  • Updated sensors

    • Add nearby ants

    • Blob sensors

  • Various engine additions

    • Set up environment by hand

    • Run multiple experiments in parallel (in progress)

    • … Starvation


Questions
Questions?

Funny ant stories?


ad