Federated Learning for End-to-End Naturalistic Driving: Generalization, Stability, and Knowledge Transfer

García Fernández, ManuelJuárez Molera, MaríaAliane, NourdineFernández Andrés, Javier2026-10-082026-10-082026Garcia-Fernandez, M., Molera, M. J., Aliane, N., & Fernandez-Andres, J. (2026). Federated learning for end-to-end naturalistic driving: Generalization, stability, and knowledge transfer. IEEE Access, 14, 151612-151626. https://doi.org/10.1109/ACCESS.2026.37389762169-3536https://hdl.handle.net/11268/17466This paper presents a systematic evaluation of generalization and knowledge-sharing capabilities in end-to-end naturalistic driving under Federated Learning. Using the Autonomous Naturalistic obstacle-Avoidance Interaction Dataset dataset, comprising four heterogeneous drivers performing real-world steering and obstacle-avoidance maneuvers, PilotNet is assessed across three training paradigms: isolated local training, centralized training on pooled data, and horizontal FL using FedAvg and FedProx aggregation protocols.engAttribution 4.0 Internationalhttp://creativecommons.org/licenses/by/4.0/Computación y tecnologíaFederated Learning for End-to-End Naturalistic Driving: Generalization, Stability, and Knowledge Transferjournal article10.1109/ACCESS.2026.3738976open accessIngeniería del transporteVehículo automotorIngeniería mecánicaGoal 9: Build resilient infrastructure, promote sustainable industrialization and foster innovation