How to use Evolve-A-Robot

The information in this modal box explains how to set up and run a simulation. It is divided into two sections, “Settings Menu” and “Window Controls”.

Settings Menu

The sections below correspond to the menu selections available on the left-hand side of the window. The default settings have been chosen to make an “interesting” simulation, but feel free to experiment with the settings to see what effect they have on the evolved robot.

Evolution

• The first slider sets the population size (default 20). Population size can affect the rate at which the simulation converges and the genetic diversity.
• Mutation settings affect how quickly mutations are introduced into the population (Rate) and what method is used to generate mutation (Random or Gaussian).
• Crossover refers to the algorithm that is used for generating offspring. It can be changed by manipulating the Method (Tournament, Roulette, Random) and the Rate.
  • If Tournament is selected there is also the option available to change the number of individuals involved in the tournament.
  • The crossover Rate can be changed from 0 to 100 percent.
• Elitism (making sure the best individual remains a member of the population in the next generation) can be turned on and off using the switch.

Environment

• Choose from four different environments:

  • Flat
  • Stairs
  • Logs
  • Spheres

  Each environment has its own set of parameters to control. Choose and experiment!

Visualization

• The first option is for choosing how many individuals in a population to render at the same time. Must be an integer between 1 and the population size.
• The Graphing switch toggles the fitness graph.

Window Controls

After the parameters have been chosen in the Settings Menu, click the “Play/Pause” button in the main window to start the simulation. The same button will pause the simulation at any time. Click the “Reset” button to restart the simulation with a new population.

All about…

Evolve-A-Robot

Evolve-A-Robot was developed as part of a summer Research Experience for Teachers (RET, see below) and is intended to be an instructional aid for teaching Evolutionary Computation (EC). It can be used equally well by students new to the concepts of EC as well as students that have already had an introduction to Genetic Algorithms (GA). It requires only that the viewing browser has WebGL enabled (Chrome recommended; enabled by default).

The RET Program

This website is based upon work supported by the National Science Foundation under Grant No. CNS 1300794 administered by the College of Engineering at Michigan State University. Click here for more information.

The Authors

Jared M. Moore

Jared started this project on his own website (Jared M Moore) as a means for teaching evolutionary robotics to K-12 students. He quickly discovered, however, that Evolve-A-Robot could be of use to the ALIFE community in general.

Anthony J. Clark

Like Jared, Tony has used WebGL technologies in his own research, and joined the project to help make Evolve-A-Robot a better teaching tool.

Charles M. Payson

Charlie is a mechanical engineer and award winning teacher at the Battle Creek Area Math and Science Center in Michigan. Through the RET program, Charlie works with Jared and Tony on summer research projects relating to evolutinoary robotics.

Inspiration

It would be remiss of us not to mention boxcar2d.com. A good site that can be used as an introduction to or in parallel with this one.