User-Adaptive Swarm Interface through Eye-Tracking

Robotics Master Thesis - EPFL

• Eye-tracking models for cognitive load estimation of a user
• Data analysis pipeline

Installation

I suggest using the python uv (see here) package manager as it allows to install speciific python versions easily.

1. Clone the git repo

git clone https://github.com/Alex0021/thesis-adaptive-swarm-interfaces.git 

2. Create the venv inside the workload_inference service

cd services/workload_inference
uv venv .venv --python=3.14
./.venv/Scripts/activate

Note: the python version is specified as 3.14 here. The activation script might be under a slightly diffferent location for non windows users

3. Install the required packages

uv sync
uv pip install -e .

Note: Make sure to use the -e (editable) mode, otherwise the relative path resolution won't work (should be improved at some point)

Usage

The workload_inference command will launch the different experiment interfaces

workload_inference --experiment [nback|gates]

The plot_results is used to generate the different plots for analysis

plot_results --help

Other scripts are also available, just have a look at the pyproject.toml file