Jérôme Pasquet
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Jérôme Pasquet has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Zamirddine Mari
DGA Techniques Navales—Direction Générale de l’Armement, 83000 Toulon, France
2 shared publications Agnès Rico
1 shared publicationBertrand Limier
ISEM Univ Montpellier, CNRS, IRD 34090 Montpellier France
1 shared publication- Betty Gimenez
ISEM Univ Montpellier, CNRS, IRD 34090 Montpellier France
1 shared publication - Céline Devaux
University of Montpellier, CNRS, IRD, ISEM, Montpellier, France
1 shared publication - Jean-Frédéric Terral
ISEM, Institut des Sciences de l’Evolution-Montpellier, Univ Montpellier, CNRS, IRD, EPHE , Montpellier
1 shared publication
Latest Publications
- Open Access
Hierarchical Two-Agent Reinforcement Learning for Long-Horizon Energy Optimization and Short-Horizon Navigation in Autonomous Electric Boats
This paper presents a hierarchical two-agent deep reinforcement learning architecture for an electric racing boat, addressing the joint challenges of long-horizon energy management and short-horizon autonomous navigation. The system integrates two independently trained agents: an energy optimization agent based on an importance-weighted actor–learner architecture, selected for its ability to handle long-horizon decision-making and distributed training, and a navigation agent trained with Proximal Policy Optimization, chosen for its training stability. The navigation policy tracks a target speed issued by the energy agent for each track segment. This target speed is the only coupling between the two levels. A curriculum learning strategy is used for both tasks. In a race simulator calibrated on real telemetry, the energy agent reaches the full 34-lap configuration, with a training success rate of 0.51 on average and about 0.72 at the end of training, while the navigation agent reaches its 0.80 evaluation success criterion under a ±5 km/h tolerance. A Unity3D-based environment with hydrodynamic modeling and autopilot software-in-the-loop emulation provides a framework for validation prior to on-water deployment; the closed-loop evaluation of the coupled agents remains future work.
Open AccessHierarchical Two-Agent Reinforcement Learning for Long-Horizon Energy Optimization and Short-Horizon Navigation in Autonomous Electric Boats
This paper presents a hierarchical two-agent deep reinforcement learning architecture for an electric racing boat, addressing the joint challenges of long-horizon energy management and short-horizon autonomous navigation. The system integrates two independently trained agents: an energy optimization agent based on an importance-weighted actor–learner architecture, selected for its ability to handle long-horizon decision-making and distributed training, and a navigation agent trained with Proximal Policy Optimization, chosen for its training stability. The navigation policy tracks a target speed issued by the energy agent for each track segment. This target speed is the only coupling between the two levels. A curriculum learning strategy is used for both tasks. In a race simulator calibrated on real telemetry, the energy agent reaches the full 34-lap configuration, with a training success rate of 0.51 on average and about 0.72 at the end of training, while the navigation agent reaches its 0.80 evaluation success criterion under a ±5 km/h tolerance. A Unity3D-based environment with hydrodynamic modeling and autopilot software-in-the-loop emulation provides a framework for validation prior to on-water deployment; the closed-loop evaluation of the coupled agents remains future work.
- Open Access
- 18 July 2026
- Published by Elsevier in SSRN Electronic Journal
Interval-valued prediction: the MacSum neural network
Accurate prediction is a prime aim of regression models, particularly in applications where decision making depends on the predicted value and on an assessment of its reliability. Neural networks have demonstrated remarkable performance as universal function approximators, yet they generally provide point predictions without explicitly quantifying the associated uncertainty. Current uncertainty estimation approaches mainly rely on prediction intervals, Bayesian inference, or ensemble and bootstrapping techniques, often at the cost of increased computational complexity or training time.Here we propose a novel neural network architecture that directly predicts output intervals while preserving the simplicity and flexibility of conventional deep learning frameworks. The proposed strategy replaces conventional linear aggregation performed within each neuron by a non-linear interval-valued aggregation operator, referred to as MacSum, followed by standard activation functions. This formulation enables the network to naturally propagate interval information through its layers and generate prediction intervals rather than single-valued outputs. Unlike many interval-based or probabilistic methods, the proposed model is compatible with gradient-based optimization, thereby allowing efficient training using standard backpropagation techniques.The framework we propose here aims to capture both data-related and model-related uncertainties, while maintaining computational complexity comparable to that of conventional neural networks. The method combines interval analysis with deep learning principles, thus offering an alternative pathway to uncertainty-aware prediction that can be readily integrated into existing machine learning architectures. The approach is particularly relevant for safety-critical applications where reliable confidence estimates are essential alongside accurate predictions.
- 18 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessInterval-valued prediction: the MacSum neural network
Accurate prediction is a prime aim of regression models, particularly in applications where decision making depends on the predicted value and on an assessment of its reliability. Neural networks have demonstrated remarkable performance as universal function approximators, yet they generally provide point predictions without explicitly quantifying the associated uncertainty. Current uncertainty estimation approaches mainly rely on prediction intervals, Bayesian inference, or ensemble and bootstrapping techniques, often at the cost of increased computational complexity or training time.Here we propose a novel neural network architecture that directly predicts output intervals while preserving the simplicity and flexibility of conventional deep learning frameworks. The proposed strategy replaces conventional linear aggregation performed within each neuron by a non-linear interval-valued aggregation operator, referred to as MacSum, followed by standard activation functions. This formulation enables the network to naturally propagate interval information through its layers and generate prediction intervals rather than single-valued outputs. Unlike many interval-based or probabilistic methods, the proposed model is compatible with gradient-based optimization, thereby allowing efficient training using standard backpropagation techniques.The framework we propose here aims to capture both data-related and model-related uncertainties, while maintaining computational complexity comparable to that of conventional neural networks. The method combines interval analysis with deep learning principles, thus offering an alternative pathway to uncertainty-aware prediction that can be readily integrated into existing machine learning architectures. The approach is particularly relevant for safety-critical applications where reliable confidence estimates are essential alongside accurate predictions.
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Modélisation, détection et classification d'objets urbains à partir d’images photographiques aériennes
Cette thèse aborde des problèmes liés à la localisation et reconnaissance d'objets urbains dans des images aériennes de très haute définition. Les objets urbains se caractérisent par une représentation très variable en terme de forme, texture et couleur. De plus, ils sont présents de multiples fois sur les images à analyser et peuvent être collés les uns aux autres. Pour effectuer la localisation et reconnaissance automatiquement des différents objets nous proposons d'utiliser des approches d'apprentissage supervisé. De part leurs caractéristiques, les objets urbains sont difficilement détectables et les approches classiques de détections n'offrent pas de performances satisfaisantes. Nous avons proposé l'utilisation d'un réseau de séparateurs à vaste marge (SVM) afin de mieux fusionner les informations issues des différentes résolutions et donc d'améliorer la représentativité de l'objet urbain. L'utilisation de réseau de SVM permet d'améliorer les performances mais à un coût calculatoire important. Nous avons alors proposé d'utiliser un chemin d'activation permettant de réduire la complexité sans perdre en efficacité. Ce chemin va activer le réseau de manière séquentielle et stoppera l'exploration lorsque la probabilité de détection d'un objet est importante. Dans le cas d'une localisation basée sur l'extraction de caractéristiques puis la classification, la réduction calculatoire est d'un facteur cinq. Par la suite, nous avons montré que nous pouvons combiner le réseau de SVM avec les cartes de caractéristiques issues de réseaux de neurones convolutifs. Cette architecture combinée avec le chemin d'activation permet une réduction théorique du coût d'activation pouvant aller jusqu'à 97% avec un gain de performances d'environ 8% sur les données utilisées. Les méthodes développées ont pour objectif d'être intégrées dans un logiciel de la société Berger-Levrault afin de faciliter et d'améliorer la gestion de cadastre dans les collectivités locales.
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Modélisation, détection et classification d'objets urbains à partir d’images photographiques aériennes
Cette thèse aborde des problèmes liés à la localisation et reconnaissance d'objets urbains dans des images aériennes de très haute définition. Les objets urbains se caractérisent par une représentation très variable en terme de forme, texture et couleur. De plus, ils sont présents de multiples fois sur les images à analyser et peuvent être collés les uns aux autres. Pour effectuer la localisation et reconnaissance automatiquement des différents objets nous proposons d'utiliser des approches d'apprentissage supervisé. De part leurs caractéristiques, les objets urbains sont difficilement détectables et les approches classiques de détections n'offrent pas de performances satisfaisantes. Nous avons proposé l'utilisation d'un réseau de séparateurs à vaste marge (SVM) afin de mieux fusionner les informations issues des différentes résolutions et donc d'améliorer la représentativité de l'objet urbain. L'utilisation de réseau de SVM permet d'améliorer les performances mais à un coût calculatoire important. Nous avons alors proposé d'utiliser un chemin d'activation permettant de réduire la complexité sans perdre en efficacité. Ce chemin va activer le réseau de manière séquentielle et stoppera l'exploration lorsque la probabilité de détection d'un objet est importante. Dans le cas d'une localisation basée sur l'extraction de caractéristiques puis la classification, la réduction calculatoire est d'un facteur cinq. Par la suite, nous avons montré que nous pouvons combiner le réseau de SVM avec les cartes de caractéristiques issues de réseaux de neurones convolutifs. Cette architecture combinée avec le chemin d'activation permet une réduction théorique du coût d'activation pouvant aller jusqu'à 97% avec un gain de performances d'environ 8% sur les données utilisées. Les méthodes développées ont pour objectif d'être intégrées dans un logiciel de la société Berger-Levrault afin de faciliter et d'améliorer la gestion de cadastre dans les collectivités locales.
- Open Access
Curriculum-Based Reinforcement Learning for Autonomous UAV Navigation in Unknown Curved Tubular Conduits
Autonomous drone navigation in confined tubular environments remains a major challenge due to the constraining geometry of the conduits, the proximity of the walls, and the perceptual limitations inherent to such scenarios. We propose a reinforcement learning (RL) approach enabling a drone to navigate unknown three-dimensional tubes without any prior knowledge of their geometry, relying solely on local observations from a Light Detection and Ranging (LiDAR) sensor and a conditional visual detection of the tube center. In contrast, the Pure Pursuit algorithm, used as a deterministic baseline, benefits from explicit access to the centerline, creating an information asymmetry designed to assess the ability of RL to compensate for the absence of a geometric model. The agent is trained through a progressive curriculum learning strategy that gradually exposes it to increasingly curved geometries, where the tube center frequently disappears from the visual field. A turning-negotiation mechanism, based on the combination of direct visibility, directional memory, and LiDAR symmetry cues, proves essential for ensuring stable navigation under such partial observability conditions. Experiments show that the Proximal Policy Optimization (PPO) policy acquires robust and generalizable behavior, consistently outperforming the deterministic controller despite its limited access to geometric information. Validation in a high-fidelity three-dimensional environment further confirms the transferability of the learned behavior to continuous physical dynamics. In particular, this work introduces an explicit formulation of the turn negotiation problem in tubular navigation, coupled with a reward design and evaluation metrics that make turn-handling behavior measurable and analyzable. This explicit focus on turn negotiation distinguishes our approach from prior learning-based and heuristic methods. The proposed approach thus provides a complete framework for autonomous navigation in unknown tubular environments and opens perspectives for industrial, underground, or medical applications where progressing through narrow and weakly perceptive conduits represents a central challenge.
Open AccessCurriculum-Based Reinforcement Learning for Autonomous UAV Navigation in Unknown Curved Tubular Conduits
Autonomous drone navigation in confined tubular environments remains a major challenge due to the constraining geometry of the conduits, the proximity of the walls, and the perceptual limitations inherent to such scenarios. We propose a reinforcement learning (RL) approach enabling a drone to navigate unknown three-dimensional tubes without any prior knowledge of their geometry, relying solely on local observations from a Light Detection and Ranging (LiDAR) sensor and a conditional visual detection of the tube center. In contrast, the Pure Pursuit algorithm, used as a deterministic baseline, benefits from explicit access to the centerline, creating an information asymmetry designed to assess the ability of RL to compensate for the absence of a geometric model. The agent is trained through a progressive curriculum learning strategy that gradually exposes it to increasingly curved geometries, where the tube center frequently disappears from the visual field. A turning-negotiation mechanism, based on the combination of direct visibility, directional memory, and LiDAR symmetry cues, proves essential for ensuring stable navigation under such partial observability conditions. Experiments show that the Proximal Policy Optimization (PPO) policy acquires robust and generalizable behavior, consistently outperforming the deterministic controller despite its limited access to geometric information. Validation in a high-fidelity three-dimensional environment further confirms the transferability of the learned behavior to continuous physical dynamics. In particular, this work introduces an explicit formulation of the turn negotiation problem in tubular navigation, coupled with a reward design and evaluation metrics that make turn-handling behavior measurable and analyzable. This explicit focus on turn negotiation distinguishes our approach from prior learning-based and heuristic methods. The proposed approach thus provides a complete framework for autonomous navigation in unknown tubular environments and opens perspectives for industrial, underground, or medical applications where progressing through narrow and weakly perceptive conduits represents a central challenge.
- Open Access
- 28 May 2024
- Published by Wiley in New Phytologist
- Vol. 243 (2) , 797-810
A user‐friendly method to get automated pollen analysis from environmental samples
Summary: Automated pollen analysis is not yet efficient on environmental samples containing many pollen taxa and debris, which are typical in most pollen‐based studies. Contrary to classification, detection remains overlooked although it is the first step from which errors can propagate. Here, we investigated a simple but efficient method to automate pollen detection for environmental samples, optimizing workload and performance. We applied the YOLOv5 algorithm on samples containing debris and c. 40 Mediterranean plant taxa, designed and tested several strategies for annotation, and analyzed variation in detection errors. About 5% of pollen grains were left undetected, while 5% of debris were falsely detected as pollen. Undetected pollen was mainly in poor‐quality images, or of rare and irregular morphology. Pollen detection remained effective when applied to samples never seen by the algorithm, and was not improved by spending time to provide taxonomic details. Pollen detection of a single model taxon reduced annotation workload, but was only efficient for morphologically differentiated taxa. We offer guidelines to plant scientists to analyze automatically any pollen sample, providing sound criteria to apply for detection while using common and user‐friendly tools. Our method contributes to enhance the efficiency and replicability of pollen‐based studies.
- 28 May 2024
- Published by Wiley in New Phytologist
- Vol. 243 (2) , 797-810
Open AccessA user‐friendly method to get automated pollen analysis from environmental samples
Summary: Automated pollen analysis is not yet efficient on environmental samples containing many pollen taxa and debris, which are typical in most pollen‐based studies. Contrary to classification, detection remains overlooked although it is the first step from which errors can propagate. Here, we investigated a simple but efficient method to automate pollen detection for environmental samples, optimizing workload and performance. We applied the YOLOv5 algorithm on samples containing debris and c. 40 Mediterranean plant taxa, designed and tested several strategies for annotation, and analyzed variation in detection errors. About 5% of pollen grains were left undetected, while 5% of debris were falsely detected as pollen. Undetected pollen was mainly in poor‐quality images, or of rare and irregular morphology. Pollen detection remained effective when applied to samples never seen by the algorithm, and was not improved by spending time to provide taxonomic details. Pollen detection of a single model taxon reduced annotation workload, but was only efficient for morphologically differentiated taxa. We offer guidelines to plant scientists to analyze automatically any pollen sample, providing sound criteria to apply for detection while using common and user‐friendly tools. Our method contributes to enhance the efficiency and replicability of pollen‐based studies.