https://github.com/rodrigosetti/springbots

Genetic algorithm experiment
https://github.com/rodrigosetti/springbots

evolution-simulation evolutionary-algorithms genetic-algorithm

Last synced: 10 months ago
JSON representation

Genetic algorithm experiment

• Host: GitHub
• URL: https://github.com/rodrigosetti/springbots
• Owner: rodrigosetti
• License: other
• Created: 2011-04-12T23:53:04.000Z (almost 15 years ago)
• Default Branch: master
• Last Pushed: 2020-05-24T23:02:18.000Z (over 5 years ago)
• Last Synced: 2025-02-08T20:47:38.333Z (12 months ago)
• Topics: evolution-simulation, evolutionary-algorithms, genetic-algorithm
• Language: Python
• Homepage:
• Size: 162 KB
• Stars: 1
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Springbots

http://www.youtube.com/watch?v=U4vJP5q3jlo

## What is it?

Springbot is a genetic algorithm experiment in which physical simulated 2d

creatures built with nodes and springs evolve to walk, swim, jump or whatever

fitness function the user decides.  After several simulations it's possible

for one to see animal-like behaviors optimized for the desired fitness.

## Requirements

 * Python >= 2.5

 * python-qt4 (optional: springbots editor)

 * xmlrpclib-python (optional: network distribution)

 * pygame (optional: real time graphics)

## Scripts

 * `evolve.py` - Take an population described in a xml file in the stardard

                 input and evolve them for a specific fitness function chosen in

                 command line (see --help), writing statistics in standard

                 output and saving the population snapshots every several

                 iterations. Optionaly it can be run with real time graphics

                 using pygame.

 * `net-evolve.py` - Same like evolve.py but connect to a set of servers listed

                     in `fitness-servers.txt` to do the processing.

 * `fitness-server.py` - Start a fitness server, optionaly with real time

                         graphics. Listens to a TCP port accepting fitness

                         tests requests and returning the fitness itself.

 * `imagebot.py` - Transform a xml springbot description into a svg image.

 * `randombot.py` - Generate a random springbot population xml file writting in

                    standard outpuy, useful for a starting seed for

                    simulations.

 * `viewer.py` - Real time graphics of the springbot being simulated. does not

                 evolve, just for viewing purpouse.

 * `demo.py` - Takes a population xml as standard input and shows each springbot

               for a specific amount of time on it adapted environment, optionally

               includes some randoms too. This program was designed for

               demonstration purpouses.

 * `log-plot.py` - Plots a statistic graph from evolution experiments output

                   based on very flexible parameters.

## Examples

```console

# Evolve a starting random population of 100 genomes with graphics and

# statistic output:

$ ./randombot.py -p 100 | ./evolve.py -f swim -vg

# To start editor move to editor directory and run editor.py:

$ ./editor.py

# Depending on your system configuration you can run the editor direct from

# file manager.

```