ACAhmed Nait Chabane

Ahmed Nait Chabane

*This is an algorithmically generated author record based on publicly available information.

Ahmed Nait Chabane has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Unified Sparse Diffusion Transformer for Pulse-Based Lithium-Ion Battery State-of-Health Estimation and Circular-Economy Decision Support

    by Mohamed Amin Benatia, Ahmed Nait Chabane, Meriem Hafsi

    Accurate lithium-ion battery state-of-health (SoH) estimation is essential for second-life screening, predictive maintenance, and circular-economy routing decisions such as reuse, remanufacture, or recycling. Short pulse-voltage tests provide a fast diagnostic alternative to full capacity tests, but their interpretation remains challenging because pulse responses depend jointly on chemistry, state of charge (SOC), nominal capacity, degradation history, and operating protocol. This challenge becomes especially critical when a model is deployed at SOC conditions that were not observed during training.This paper proposes the \emph{Unified Sparse Diffusion Transformer} (USDT), a chemistry-conditioned sequence architecture for pulse-based battery SoH estimation and second-life decision support. USDT represents the pulse response as a sequence of voltage tokens enriched with SOC, capacity, pulse-width, and chemistry information. The architecture combines chemistry-conditioned token embedding, lightweight sequence mixing, sparse expert routing, decoupled temporal modeling, and a decision-oriented SoH regression head. An enhanced USDTPlus variant further exposes voltage-shape information through local difference channels, attention pooling, and tabular skip connections.Experiments are conducted on the processed PulseBat dataset, comprising 270 retired lithium-ion cells across LFP, LMO, and NMC chemistries. We evaluate three complementary regimes: standard battery-disjoint interpolation, seen-SOC testing, and unseen-SOC generalization. The obtained results shows that under the standard battery-disjoint interpolation protocol, strong tabular baselines remain highly competitive, with RBF-SVR achieving the best standalone regression accuracy. However, under the stricter unseen-SOC protocol, where models are trained on SOC values of 5, 25, and 50\% and tested on previously unseen SOC values, USDT-family models clearly outperform the tabular baselines. USDTPlus achieves the best unseen-SOC performance, reducing MAE by 42.5\% relative to RBF-SVR and improving $R^2$ from 0.7215 to 0.9193.Beyond regression accuracy, we evaluate reuse-threshold decision quality at $\mathrm{SoH}\geq0.80$ and quantify retained reusable energy capacity and avoided CO$_2$ under scenario-based environmental assumptions. In the unseen-SOC setting, USDTPlus substantially reduces false rejects compared with RBF-SVR, preserving more viable cells for second-life use. Under the central environmental scenario, USDTPlus increases retained reusable energy capacity and net avoided CO$_2$ relative to the strongest tabular baseline. These results show that the value of chemistry-conditioned sequence modeling is most evident when SoH estimation is used not only as a fixed-split regression task, but as a deployment-oriented decision tool for circular battery management.

  • Open Access

    Energy-aware Bi-objective Multi-robot Inspection Planning with Partial Recharging and Speed Optimization

    by Germain Junior Avossevou, Ouahib Guenounou, Ahmed Nait Chabane, Karim Beddiar

    Autonomous industrial inspection increasingly relies on heterogeneous robotic fleets operating under strict energy constraints. Efficient coordination of such systems requires the joint optimization of task allocation, routing, motion planning, and battery management under a trade-off between mission duration and energy consumption. In this work, we propose a multi-objective optimization framework for energy-aware multi-robot inspection in industrial environments. The problem is formulated as a bi-objective model that minimizes both the mission makespan and the total energy consumption. The proposed model explicitly accounts for heterogeneous robot sensing capabilities, battery capacity constraints, partial recharging strategies, and continuous velocity-profile parameters that significantly affect motion-related energy consumption. To solve this problem, a tailored NSGA-II algorithm based on a hybrid discrete--continuous encoding is developed, with adapted crossover and mutation operators to jointly optimize task allocation, routing sequences, and velocity parameters. Simulation experiments conducted on synthetic benchmark instances demonstrate the impact of speed-profile selection and recharging strategies on mission performance. The results show that jointly optimizing allocation, routing, and motion parameters yields well-distributed Pareto-optimal solutions, revealing significant trade-offs between time efficiency and energy savings. Overall, the proposed framework provides a unified approach for integrating motion- and task-level energy modeling into large-scale multi-robot inspection planning, thereby contributing to safer and more energy-efficient industrial automation.

  • Open Access

    Scalable Sparse Diffusion Transformer for Lithium-Ion Battery State-of-Health Estimation under Pulse-Testing Protocols

    by Mohamed Amin Benatia, Ahmed Nait Chabane, Meriem Hafsi

    Accurate estimation of lithium-ion battery state-of-health (SoH) is essential for predictive maintenance and second-life screening. Pulse-testing protocols capture informative short-term dynamics, yet learning the coupling between long-term degradation and transient electrochemical responses remains challenging. Moreover, many deep learning methods either incur quadratic attention cost with sequence length or generalize poorly across heterogeneous chemistries and operating regimes.This paper proposes a Scalable Sparse Diffusion Transformer (SSDT) for SoH estimation from pulse-test data. SSDT integrates (i) shallow feed-forward attention to mitigate quadratic attention overhead, (ii) sparse Mixture-of-Experts routing to increase model capacity with controlled compute, (iii) a decoupled temporal module that separates slow degradation trends from fast pulse dynamics, and (iv) a single-stage latent diffusion denoising backbone to improve representation robustness.Experiments on the PulseBat dataset, which includes heterogeneous lithium-ion cells spanning multiple chemistries, show that SSDT attains an R² of 0.94 for SoH prediction and provides improved accuracy compared with commonly used recurrent and attention-based baselines. Ablation results further highlight the benefits of sparse routing and temporal decoupling for the accuracy–runtime trade-off. Overall, SSDT offers a scalable and robust approach to battery health estimation that supports industrial diagnostic pipelines and second-life assessment under realistic pulse-testing conditions.

  • Segmentation invariante en rasance des images sonar latéral par une approche neuronale compétitive

    by Ahmed Nait-Chabane

    Un sonar latéral de cartographie enregistre les signaux qui ont été rétrodiffusés par le fond marin sur une large fauchée. Les signaux sont ainsi révélateurs de l’interaction entre l’onde acoustique émise et le fond de la mer pour une large plage de variation de l’angle de rasance. L’analyse des statistiques de ces signaux rétrodiffusés montre une dépendance à ces angles de rasance, ce qui pénalise fortement la segmentation des images en régions homogènes. Pour améliorer cette segmentation, l’approche classique consiste à corriger les artefacts dus à la formation de l’image sonar (géométrie d’acquisition, gains variables, etc.) en considérant un fond marin plat et en estimant des lois physiques (Lambert, Jackson, etc.) ou des modèles empiriques. L’approche choisie dans ce travail propose de diviser l’image sonar en bandes dans le sens de la portée ; la largeur de ces bandes étant suffisamment faible afin que l’analyse statistique de la rétrodiffusion puisse être considérée indépendante de l’angle de rasance. Deux types d’analyse de texture sont utilisés sur chaque bande de l’image. La première technique est basée sur l’estimation d’une matrice des cooccurrences et de différents attributs d’Haralick. Le deuxième type d’analyse est l’estimation d’attributs spectraux. La bande centrale localisée à la moitié de la portée du sonar est segmentée en premier par un réseau de neurones compétitifs basé sur l’algorithme SOFM (Self-Organizing Feature Maps) de Kohonen. Ensuite, la segmentation est réalisée successivement sur les bandes adjacentes, jusqu’aux limites basse et haute de la portée sonar. A partir des connaissances acquises sur la segmentation de cette première bande, le classifieur adapte sa segmentation aux bandes voisines. Cette nouvelle méthode de segmentation est évaluée sur des données réelles acquises par le sonar latéral Klein 5000. Les performances de segmentation de l’algorithme proposé sont comparées avec celles obtenues par des techniques classiques.

  • Open Access

    A Reproducible Engineering Workflow for Synthetic Data Generation and Sim-to-Real Evaluation in Robotic Sewer Inspection under Data Scarcity

    by Ahmed Nait Chabane, Chadi M&Apos;Sila

    Robotic inspection in confined sewer tunnels is constrained by hazardous conditions, limited accessibility, wet reflective surfaces, strong illumination variability, and severe data scarcity. These factors make representative, reproducible, and well-annotated real-world datasets costly and difficult to obtain, thereby limiting the development and evaluation of perception systems. This paper presents SewerSim, a reproducible and parameterized simulation framework implemented in NVIDIA Isaac Sim for synthetic data generation and sim-to-real evaluation in confined sewer environments. The framework provides explicit control over tunnel geometry, illumination, water state, and clutter, and generates synchronized RGB observations with pixel-accurate semantic and instance-level annotations under deterministic replay. We formalize a realistic asset-scarcity setting in which tunnel geometry is kept consistent with the real inspection environment, whereas debris assets remain generic because curated and annotated sewer-debris corpora are scarce. To evaluate the proposed workflow, we adopt a leak-resistant protocol based on sequence-level splits for synthetic data and contiguous temporal chunking of a real tunnel run. Three instance-segmentation model families, YOLO11-seg, YOLO26-seg, and Mask2Former, are used as downstream evaluation instruments under four training strategies: simulation-only, real-only, simulation pretraining followed by real fine-tuning, and joint simulation-real training. Results on a held-out real test set show that simulation pretraining followed by real fine-tuning reaches 0.62 AP50:95, whereas simulation-only training transfers only partially to geometry-dominated structures and fails to generalize to debris. Overall, SewerSim provides a reproducible engineering workflow for synthetic inspection data generation and controlled sim-to-real evaluation in visually challenging confined environments, with strong emphasis on reproducibility, transfer analysis, and engineering relevance.

  • Open Access

    Reconstruction-Based Methods for Multivariate Time Series Anomaly Detection : A Review and Taxonomy

    by Salma Errachidi, Aimilia Aimilia Tasidou, Ahmed Nait Chabane, Karim Beddiar