Multi-agents programming (IA 310) – ENSTA (till 2021)

There are 8 sessions of 3 hours:
  1. Introduciton and concept of agent
  2. From agents to multi-agent systems
  3. Reactive agents
  4. Cognitive agents: the BDI paradigm
  5. Other Cognitive agents and hybrid agents
  6. Advanced concepts 1
  7. Advanced concepts 2
  8. Project presentation
Each session consists in 1h of lecture and 2 hours of practice, except session 6. There is a homework for each of the sessions 1-5.
A one-hour exam will be held at the beginning of session 6, about the content of all first 5 sessions.
From session 6 to session 8, the practice will be dedicated to the project. At session 8, each tam will present his project
If the homework code of your homework or project does not compile, you won't get any point.
The homework are evaluated both on the code and on the theoretical part. The grade of the project will take into account the code itself, the approach and the presentation.
The grade of the project consists of the following thress elements:
  • Project: 40%
  • Exam: 30%
  • Homework: 30%
Please make sure you have access to the data-ensta gitlab. If you are planning to use your own computer, please get a recent python distribution (≥3.6) and all the tools you may need (IDE etc.) to code. If you don't know what to do, I'd strongly recommend installing Anaconda
If you have any question regarding the course you can contact me by mail . Please indicate [IA310]
The course PDF will soon be available
For your project, you can choose any topic related to multi-agent systems. The code must be easily readable, and must be in one of the languages I know:
  • python
  • Java
  • C++
  • Scala
  • Gama
Here are some examples of subjects:
  • Model of the spreading of an epidemic in a city through multi-agent system
  • An automated negotiation agent
  • Decentralized Multi-Robot Task Allocation


PhD students

MSc students

  • 2020 – 2021 Malo Simondin: « Intelligence Artificielle Collaborative / Algorithme d'allocation de tâches par enchères collaboratives tolérante aux pannes »
  • 2018 – 2019 Pierre Larrenie: « Intelligence Artificielle Collaborative / Allocation dynamique collaborative de ressources de poursuites de radars en réseaux, par des techniques d’enchères distribuées (modèle de négociation) »
  • 2017 – 2018 Douae Ahmadoun: « Etude et expérimentation d'un algorithme de coopération décentralisée dans un essaim de drones »
  • 2017 – 2018 Samuel Amoyal: « Algorithms simulations in multisensor environments »