Miniproject for BIOENG-456: Controlling behavior in animals and robots

Welcome to the Miniproject for BIOENG-456!

Setup

pynput is required for the KeyBoardController in explore_levels.py. To install it, run:

conda activate flygym
pip install pynput

Usage

To explore the levels interactively, run the explore_levels.py script:

conda activate flygym
python explore_levels.py --level <level> --seed <seed>

Replace <level> with the desired level number (0 to 4 for the 5 levels, -1 for just a flat terrain) and <seed> with the random seed for reproducibility.

The run_simulation.py script contains code that will be used to evaluate the controller.

To run the simulation, use:

conda activate flygym
python run_simulation.py <submission_folder> --level <level> --seed <seed> --progress --max-steps 10000

Replace <submission_folder> with the path to your submission folder. An example submission folder is provided in the submission directory. It contains a controller that walks forward regardless of the observations it receives. You may use that as a starting point for your own controller.

A test_controller.ipynb notebook is provided to help you test your controller. You may for example use it to get the list of observations over time and plot them when designing your controller.

Submission

The submission folder should contain the following files:

• controller.py: The controller code. It should contain a class named Controller that subclasses BaseController and implements the get_action method and done_level method.
• init.py: A file to mark the directory as a package.
• any other files you need for your controller.

You may run the check_submission.py script to check if your submission folder is valid and zip it for submission:

conda activate flygym
python check_submission.py <submission_folder>

Testing your controller

You can test your controller using the test_controller.py script, that will run it on all the levels and a selection of random seeds.

You can also test it with extra random seeds as follows

conda activate flygym
python test_controller.py <submission_folder> --extra-seeds 10,100,22

Creating a private copy while keeping track of the changes from the public repository

1. Clone this repository

git clone https://github.com/NeLy-EPFL/cobar-miniproject-2025
cd cobar-miniproject-2025

2. Create a New Private Repository on GitHub:

• Go to GitHub and create a new private repository.
• Do not initialize it with a README, .gitignore, or any other files.

3. Set the New Private Repository as a Remote:

git remote rename origin upstream
git remote add origin https://github.com/<your_username>/cobar-miniproject-2025

4. Push the Cloned Repository to Your Private Repository:

git push -u origin main

5. We will notify you if there are important changes to the repository. To fetch updates from the public repository and merge them into your private repository, use the following commands:

git fetch upstream
git merge upstream/main