Experiments with Changing Environments

This repository contains the code used in chapter 7 of Paolo Baldini's PhD Thesis. The start-up of specific instances of the experiment is managed through environmental variables set when starting the docker image.

The repository is organized as follows:

• exp contains the docker-compose files to run the two experiments
• gym contains the argos3 arenas of the two experiments
• shl contains the scripts that start the experiments in the docker container
• src contains the Lua code of the robot controllers

Finally, the Docker file is used to build the experiment image that will be pushed to a public repository:

docker build --network=host --tag quay.io/p-baldini/2025-ece:1.1.0 .
docker push quay.io/p-baldini/2025-ece:1.1.0

If you want to run or test the docker image locally you can use the following commands:

mkdir out
xhost +local:docker
docker run --rm --network=host -e DISPLAY \
    -e START_SEED=1 \
    -e END_SEED=2 \
    -e WORK_DIR=out \
    quay.io/p-baldini/2025-ece:1.1.0