Projects

MADSAR

4.0 factories are complex open and distributed systems, where equipments are dynamically able to join or quit the production line, integrate and pair themself, and communicate with both other artificial elements and human operators involved in the process. However when incidents happen, namely indicators showing measures getting too far from expected ones according to the process, being able to understand the situation and identify faulty elements might be far too difficult for operators and therefore result in a production being down for a while and important costs for the manufacturer.

Started on early 2023, this project aims to explore the possibility to assist operators in such investigations, using explainable multiagent systems (XMAS). We used yeast culture in bioreactors as a use case and built experimental setups to illustrates many key ideas and clarify remaining obstacles.

SOCRATES

SOCRATES, for SOCial Routing AgenTs Evolutionnary Simulation, is a tool designed and developped by members of the ERC BEHAVE project to explore the impact of social routing on the efficiency of transportation infrastructures. This software is designed both for academic research purpose and as a tool for policy designers. This multiagent-based simulator uses coevolutionnary algorithms to let agents explore several paths to find User Equilibrium and System Optimum according with a road network and their start and destination point on the network.

A view of SOCRATES

Nowadays (January 2021), SOCRATES is still in alpha release. Updates will be posted here as soon as possible.

Requirements

To run this software, you only need a recent Java Runtime Environment (JRE). If you do not have any yet installed, you can download some on www.oracle.com. SOCRATES has been tested only on GNU/Linux and Microsoft Windows 10, but is expected to run on any operating system having a recent JRE.

How to run an existing scenario

SOCRATES is compatible with the standards Open Street Map XML file format. You will find existing maps in the examples folder provided with the simulator. You may also download maps from OpenStreetMap.org. To load the map, click on the menu "Simulation > Open" and then select the osm file.

The population of agents can either be loaded or generated. We provide a few agents classes already functionnal, and example population files (using ".pop" as an extension in example). As for the map, to load such file you should simply click on the menu "Simulation > Open" and then select the pop file.

Credits

SOCRATES is part of the research project BEHAVE conducted in TU Delft, Faculty of Technology, Policy and Management. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 724431).

This simulator is developped in Java, using the Infonode® Docking Windows library.

Development team:

Technical details

This software uses, to make computation as fast as possible (even if the visual render looks less realistic) the Equirectangular projection to display maps.

Distances are calculated using the Haversine formula.

Obfuscation-based decision making

This software designed with the jacamo framework implements a set of random, transparency-based decision making and obfuscation-based decision making agents.
Several papers and posters on various aspects of this topic are currently in submission.
Click here to see the materials (Authentification required)
Click here to download the proof of concept

A toolkit for DPA to break AES on microcontroller

Waiting first a publication of my results

Coming soon, sorry.

Ethical market simulator

An example of output

This software implemented with the JaCaMo framework is a multi-agent system used to simulate a financial market and show how a set of agents may trade assets according with ethical considerations and evaluate the behavior of the others.

Please see here how to download, install and use this software.

See also the paper Nicolas Cointe, Grégory Bonnet, Olivier Boissier, Multi-Agent Based Ethical Asset Management, Workshop on Ethics in the Design of Intelligent Agents (EDAI), ECAI, August 2016. for more information.