Maxime Toquebiau

Maxime Toquebiau

Auto-generated profile, based on publicly available information

Maxime Toquebiau

Sorbonne Université, CNRS, ISIR, F-75005 Paris, France

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 January 2025 in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
Verified
Journal Article
Open access
LCLeo Cazenille
Leo Cazenille
MTMaxime Toquebiau
Maxime Toquebiau
NLNicolas Lobato-Dauzier
Nicolas Lobato-Dauzier
ALAlessia Loi
Alessia Loi
LMLoona Macabre
Loona Macabre
NBNicolas Bredeche
Nicolas Bredeche
This paper investigates the role of communication in improving coordination within robot swarms, focusing on a paradigm where learning and execution occur simultaneously in a decentralized manner. We highlight the role communication can play in addressing the credit assignment problem (individual contribution to the overall performance), and how it can be influenced by it. We propose a taxonomy of existing and future works on communication, focusing on information selection and physical abstraction as principal axes for classification: from low-level lossless compression with raw signal extraction and processing to high-level lossy compression with structured communication models. The paper reviews current research from evolutionary robotics, multi-agent (deep) reinforcement learning, language models and biophysics models to outline the challenges and opportunities of communication in a collective of robots that continuously learn from one another through local message exchanges, illustrating a form of social learning. This article is part of the theme issue ‘The road forward with swarm systems’.

21 October 2025
Verified
Book Chapter
Open access
MTMaxime Toquebiau
Maxime Toquebiau
JKJae Yun JUN KIM
Jae Yun JUN KIM
FBFaïz Benamar
Faïz Benamar
NBNicolas Bredeche
Nicolas Bredeche
Most prior works on communication in multi-agent reinforcement learning have focused on emergent communication, which often results in inefficient and non-interpretable systems. Inspired by the role of language in natural intelligence, we investigate how grounding agents in a human-defined language can improve the learning and coordination of embodied agents. We propose a framework in which agents are trained not only to act but also to produce and interpret natural language descriptions of their observations. This language-augmented learning serves a dual role: enabling efficient and interpretable communication between agents, and guiding representation learning. We demonstrate that language-augmented agents outperform emergent communication baselines across various tasks. Our analysis reveals that language grounding leads to more informative internal representations, better generalization to new partners, and improved capability for human-agent interaction. These findings demonstrate the effectiveness of integrating structured language into multi-agent learning and open avenues for more interpretable and capable multi-agent systems.1 1 Appendices and code are available at github.com/MToquebiau/LAMARL

9 April 2026
Verified
Thesis
Closed access
MTMaxime Toquebiau
Maxime Toquebiau
Apprentissage par renforcement profond pour les systèmes multi-agents appliqué à la robotique mobile: L'apprentissage par renforcement profond permet l'apprentissage de stratégies multi-agents dans divers contextes. Malgré des avancées significatives au cours de la dernière décennie, ils rencontrent encore d'importants défis pour l'apprentissage des compétences nécessaires dans les contextes robotiques. Ce travail examine l'état actuel du MADRL, identifie certaines de ses lacunes et propose des méthodes innovantes pour améliorer les approches existantes. La thèse se concentre sur deux défis critiques : l'exploration et la communication. Les approches d'apprentissage multi-agent reconnues ont du mal à découvrir des comportements conjoints optimaux en raison de stratégies d'exploration locales inefficaces. Pour résoudre ce problème, nous introduisons une méthode basée sur la motivation intrinsèque qui encourage les agents à explorer de nouvelles actions coordonnées, améliorant ainsi la convergence vers des stratégies conjointes optimales. Dans le cas de l'apprentissage de la communication, bien que la communication émergente différenciable soit une approche populaire, nous soutenons qu'elle pourrait ne pas répondre pleinement aux besoins des agents artificiels, notamment pour les applications robotiques. À la place, nous proposons d'enseigner une langue prédéfinie aux agents. Ceci permet d'acquérir de multiples avantages : communication efficace, apprentissage de représentation ancré dans le langage, généralisation améliorée et interaction plus facile avec les êtres humains. Nous proposons un algorithme d'apprentissage pour enseigner un langage à des agents et présentons des expériences démontrant l'efficacité de cette approche.

6 February 2024 in arXiv
Verified
Preprint
Open access
PDF file
MTMaxime Toquebiau
Maxime Toquebiau
NBNicolas Bredeche
Nicolas Bredeche
FBFaïz Benamar
Faïz Benamar
JKJae Yun JUN KIM
Jae Yun JUN KIM
Multi-agent deep reinforcement learning (MADRL) problems often encounter the challenge of sparse rewards. This challenge becomes even more pronounced when coordination among agents is necessary. As performance depends not only on one agent's behavior but rather on the joint behavior of multiple agents, finding an adequate solution becomes significantly harder. In this context, a group of agents can benefit from actively exploring different joint strategies in order to determine the most efficient one. In this paper, we propose an approach for rewarding strategies where agents collectively exhibit novel behaviors. We present JIM (Joint Intrinsic Motivation), a multi-agent intrinsic motivation method that follows the centralized learning with decentralized execution paradigm. JIM rewards joint trajectories based on a centralized measure of novelty designed to function in continuous environments. We demonstrate the strengths of this approach both in a synthetic environment designed to reveal shortcomings of state-of-the-art MADRL methods, and in simulated robotic tasks. Results show that joint exploration is crucial for solving tasks where the optimal strategy requires a high level of coordination.