Milo's Softrobot Project (adapted from Yet Another Soft Robot Evolver)

This repository uses Evolution Gym. It's a project in my short-term exchange period in University of Tsukuba.

Files

• Search.py: uses simple search algorithms to optimize the robot body for a given task. Random Search, GA and ES are implemented. Use python Search.py -h for options.

• Visualize.py: Visualizes the result of one robot running on one world file (note: it does not need to be the same ones that ran together originally!). Use python Visualize -h for options.

• run.sh: Used to run batch processes with one command. $./run.sh

• EC_algorithms/: To store algorithms and functions used by algorithms.

• EC_algorithms/utils.py: Functions used by algorithms.

• EC_algorithms/logger.py: A object used for record information during the algorithm running.

• world/: Objects define calculation method or start position for different task. The world json files store in evogym_world folder. sim_files include the world files in Evogym that not transform yet. T_world folder include the world files compatible with YASRE.

• world/json_transformer.py: A function can transform world in Evogym to a json file can be used in YASRE.

• robot/: Contain objects for robot.

• log/: Store log files. You can change the path with -d option in Search.py.

• evaluation/evaluate.py: Define evaluation method.

How to Install:

• Create a local python 3.10 environment using pyenv or similar
• Install evogym: pip install evogym --upgrade (See evogym repository for details)
• Install linux dependencies (for evogym): sudo apt install xorg-dev libglu1-mesa-dev
• Install python dependencies (for evogym): pip install glfw PyOpenGL ttkbootstrap
• Install python dependencies (for this repo): pip install pygifsicle imageio

Guide

The beginner's guide