Home » Teams » Amac » Nicolas Fontbonne
  • Nicolas Fontbonne

  • PhD student
  • Team: Amac
  • Office: 310
  • Email: nicolas.fontbonne@sorbonne-universite.fr
  • Website: https://www.linkedin.com/in/nicolas-fontbonne/
  • Bio: Ph.D. student supervised by Nicolas Bredeche (ISIR, SU) and Nicolas Maudet (SMA, LIP6) on the topic Cooperative learning and individual contribution estimation for adaptive swarm robotics.
    My research focuses on cooperation between robots in a context where they have to learn their control policies autonomously. These policies are optimized with machine learning algorithms that take advantage of a reward function to increase performance progressively. The structure of this function will significantly influence the learning dynamics and, thus, the possible behaviours of the agents.
    The first axis concerns systems where the agents individually receive a local reward adapted to their actions and must converge toward an optimal collective behaviour. In this context, I study Embodied Evolution algorithms, and in particular the Horizontal Information Transfer algorithm which allows online distributed robot swarm optimization.
    The second axis concerns systems where the reward is given globally to the whole team. Therefore, this evaluation does not necessarily represent each agent's performance, and it can be challenging to calculate an individual contribution. In this context, I study the dynamics of cooperative co-evolution algorithms (CCEA) on resource allocation and rover exploration issues. These algorithms allow an estimation of the marginal contribution and a compromise between evaluation quality and execution speed.

Publications

  • Nicolas Fontbonne, Nicolas Maudet, Nicolas Bredeche. Cooperative Co-Evolution and Adaptive Team Composition for a Multi-Rover Resource Allocation Problem. European Conference on Genetic Programming, Apr 2022, Madrid, Spain. pp.179-193, ⟨10.1007/978-3-031-02056-8_12⟩. ⟨hal-03842174⟩
  • Nicolas Bredeche, Nicolas Fontbonne. Social learning in swarm robotics. Philosophical Transactions of the Royal Society B: Biological Sciences, 2022, 377 (1843), pp.20200309. ⟨10.1098/rstb.2020.0309⟩. ⟨hal-03500739⟩
  • Paul Ecoffet, Nicolas Fontbonne, Jean-Baptiste André, Nicolas Bredeche. Policy Search with Rare Significant Events: Choosing the Right Partner to Cooperate with. PLoS ONE, 2022, PLoS ONE, 17 (4), pp.e0266841. ⟨10.1371/journal.pone.0266841⟩. ⟨hal-03315730⟩
  • Paul Ecoffet, Nicolas Fontbonne, Jean-Baptiste André, Nicolas Bredeche. Reinforcement Learning with Rare Significant Events: Direct Policy Search vs. Gradient Policy Search. Genetic and Evolutionary Computation Conference Companion, 2021, Lille (en ligne), France. ⟨hal-03315728⟩
  • Nicolas Fontbonne, Olivier Dauchot, Nicolas Bredeche. Distributed On-line Learning in Swarm Robotics with Limited Communication Bandwidth. IEEE Congress on Evolutionary Computation, 2020, Glasgow (virtual), United Kingdom. ⟨10.1109/CEC48606.2020.9185697⟩. ⟨hal-03175237⟩