A Field-based Computing Approach to Sensing-driven Clustering in Robot Swarms

This repository contains the experiments that explore sensing-driven clustering techniques applied in Aggregate Computing.

In result discussion there is a legend that explain how to interpret the graphical simulations.

Prerequisite

A working version of Java, the supported version range is 11 to 17, and a working version of Python 3, including pip.

How to launch

To run the Alchemist simulations devised from this paper, you could use the pre-defined gradle task runAllGraphic. This task will launch each simulation defined by a yaml file in src/main/yaml.

Hence, to run all the simulation you have to write

./gradlew runAllGraphic

Press P to start the simulation.

For further information about the GUI, see the graphical interface shortcuts.

In Windows, I suggest you to use WSL 2. Otherwise, you should write:

gradlew.bat runAllGraphic

Re-generating all the data

The experiment is entirely reproducible. Regenerating all the data may take weeks on a well-equipped 2021 personal computer. The process is CPU-intensive and we do not recommend running it on devices where heat can be damaging to the battery.

In order to re-run all the experiments, launch:

./gradlew runAllBatch

data will be generated into the data folder

Reproduce plots

If you are just willing to re-run the data analysis and generate all the charts in the paper (plus hundreds other), you can use the data we generated in the past:

pip install -r requirements.txt --user
./plot-all-clusters.sh