Dooders

Reality works; simulate it.

Overview

Dooders is an open-source research project focused on the development of artificial intelligent agents in a simulated reality. The project aims to enable the conditions and mechanisms for cognitive agents to evolve and emerge in a digital environment.

A Dooder is an agent object in the simulation with an amount of causal control. It acts in the simulation only as long as it consumes Energy.

Take a look on how a Dooder will learn and act, get an overview of the core components of the library, or read why I started the project.

I will also be documenting experiments in substack. Including the results from my first experiment.

The code, content, and concepts will change over time as I explore different ideas.

Everything in this repository should be considered unfinished and a work-in-progress

How to use it

from dooders import Experiment

experiment = Experiment()

experiment.simulate()

experiment.experiment_summary()


# Example output using the default settings
# This simulation ended after 53 cycle when 
# all Dooders died from starvation

{'SimulationID': 'XGZBhzoc8juERXpZjLZMPR',
 'Timestamp': '2023-03-09, 18:20:33',
 'CycleCount': 53,
 'TotalEnergy': 634,
 'ConsumedEnergy': 41,
 'StartingDooderCount': 10,
 'EndingDooderCount': 0,
 'ElapsedSeconds': 0,
 'AverageAge': 14}

For more details, see the Quick Start guide.