Research

My interests are in Artificial Intelligence (AI), and especially (deep) reinforcement learning and developmental learning and their applications to robotics (mono-robot, multi-robot and human-robot systems). More specifically, I'm interested in decisional interactions in multi-agent systems composed of a set of artificial agents and composed of artificial and human agents.

PhD/Master Students

Current Students

  • Remy Chaput (Post-Doc), co-advising with Mathieu Guillermin, Multi-agent reinforcement learning for the co- construction of ethical behaviors
  • Timon Deschamps (PhD), co-advising with Luis Gustavo Nardin and Mathieu Guillermin, Multi-objective and multi-agent reinforcement learning for the co- construction of ethical behaviors
  • Pierre Marza (PhD), co-advising with Christian Wolf and Olivier Simonin, Large-scale training of robot navigation

    • Previous Students

  • Arthur Aubret, co-advising with Salima Hassas, thesis defended on 31/11/2021, Learning increasingly complex skills through deep reinforcement learning using intrinsic motivation
  • Guillaume Bono, co-advising with Jilles Dibangoye, Olivier Simonin and Florian Pereyron, thesis defended on 28/10/2020, Deep Multi-Agent Reinforcement Learning for Dynamic and Stochastic Vehicle Routing Problems
  • Benoit Vuillemin, co-advising with Salima Hassas, Lionel Delphin-Poulat and Rozenn Nicol, thesis defended on 8/07/2020, Prediction rule mining in an Ambient Intelligence context
  • Antoine Grea, co-advising with Samir Aknine, thesis defended on 30/01/2020, Endomorphic metalanguage and abstract planning for real-time intent recognition

    • Master Students

  • Master2 (2017): Bruno Dumas, co-advising with Jilles Dibangoye and Christian Wolf
    Deep Reinforcement Learning applied to active perception of a fleet of mobile robots.
  • PFE TC INSA (2017): Hoang Vu, co-advising with Olivier Simonin
    Distributed architecture and man-machine interface for the control of a fleet of mobile robots
  • PFE GE INSA (2016): Simon Bultmann, co-advising with Olivier Simonin
    Multi-Robot navigation and cooperative mapping in a circular topology.
  • Master2 (2016): Lucas Foulon, co-advising with Frederic Armetta
    Study of the expressivity of neural networks from a constructivist perspective: application to automatic text generation.
  • Master2 (2016): Antoine Richard, co-advising with Salima Hassas and Michael Mrissa
    Self-adaptative software in Smart Environments using multi-agent based learning.
  • Master2 (2015): Remi Casado, co-advising with Amelie Cordier
    Learning of hierarchic behaviors: a comparative study between developmental and reinforcement learning.
  • Master2 (2015): Jonathan Cohen, co-advising with Olivier Simonin
    Anytime algorithms for the exploration and observation of a complex scene by a mobile multi-agent system
  • Master1 (2015): Aurelie Kong Win Chang, co-advising with Frederic Armetta
    Visualisation tools for the construction of representation in developpemental learning: application to reinforcement learning
  • PFE TC INSA (2014): Cristhian Mayor Navarro, co-advising with Olivier Simonin
    Spatial coordination and communication in a robot fleet to optimize the observation of a scene
  • Master1 (2014): Florian Bernard, Rachid Delory
    Implementation of mono and multi-robot exploration strategies with ROS



    • Multi-Robot Exploration

    Distributed Value Functions for Multi-Robot Exploration

    This work is made in collaboration with Laurent Jeanpierre and A.I. Mouaddib

    We performed experiments with our two micro-troopers. Here are some videos showing the exploration of the robots and how they spread out to different areas so as to cover the space efficiently and to minimize close interactions. Some situations of local coordination successfully resolved are also shown.

  • 1st experiment (click here to see the video)
  • 2nd experiment (click here to see the video)
  • 3rd experiment (click here to see the video)
  • 4th experiment (click here to see the video)

    These videos can be also viewed in low quality on this site: (click here)

    The video of our ROBOTS_MALINS consortium for the 2nd year of the CAROTTE challenge: (click here to see the video)



    Human-Robot Interaction

    A Model for Verbal and Non-Verbal Human-Robot Collaboration

    This work is made in collaboration with A.B. Karami and A.I. Mouaddib

    We are motivated by building a system for an autonomous robot companion that collaborates with a human partner for achieving a common mission. The objective of the robot is to infer the human's preferences upon the tasks of the mission so as to collaborate with the human by achieving human's non-favorite tasks. Inspired by recent researches about the recognition of human's intention, we propose a unified model that allows the robot to switch accurately between verbal and non-verbal interactions. Our system unifies an epistemic partially observable Markov decision process (POMDP) that is a human-robot spoken dialog system aiming at disambiguating the human's preferences and an intuitive human-robot collaboration consisting in inferring human's intention based on the observed human actions. The beliefs over human's preferences computed during the dialog are then reinforced in the course of the task execution by the intuitive interaction. Our unified model helps the robot inferring the human's preferences and deciding which tasks to perform to effectively satisfy these preferences. The robot is also able to adjust its plan rapidly in case of sudden changes in the human's preferences and to switch between both kind of interactions. Experimental results on a scenario inspired from robocup@home outline various specific behaviors of the robot during the collaborative mission.

    Experimental results

    The video (click here to see the video) shows the execution of a complete mission composed of 4 tasks following a policy computed with the approximate POMDP solver Topological Order Planner (TOP) (Dibangoye et al. 2009). The experiment uses a mobile koala robot. In order to be able to realize the verbal communication between the robot and the human, we integrated a speech recognition module for the robot to interpret the human speech utterances; and a speech synthesizer for the robot to convert its queries into speech. Audio observations are processed using the Sphinx-4 opensource speech recognition system (Walker et al. 2004) and the FreeTTS open-source speech synthesizer was used for the text-to-speech conversion (Walker et al. 2002).

    The table shows the dialog between the human and the robot illustrated in the video. Various switches between verbal and non-verbal interactions and specific behaviors of the robot during the collaboration are outlined.

    Bibliography

    Dibangoye, J. S.; Shani, G.; Chaib-draa, B.; and Mouaddib, A.-I. 2009. Topological order planner for pomdps. In IJCAI -09, 1684-1689

    W. Walker, P. Lamere, P. Kwok, B. Raj, R. Singh, E. Gouvea, P. Wolf, and J. Woelfel, "Sphinx-4: A flexible open source framework for speech recognition", Tech. Rep., 2004.

    W. Walker, P. Lamere, and P. Kwok, "Freetts - a performance case study", Tech. Rep. TR-2002-114, 2002.


    		•