Enabling Federated Object Detection for Connected Autonomous Vehicles: A Deployment-Oriented Evaluation

2025-09-01

AI Summary

This work addresses practical deployment challenges of federated learning (FL) for object detection in connected and autonomous vehicles (CAVs), tackling three key bottlenecks: non-IID data distributions across vehicles, stringent on-board computational constraints, and environmental dynamics (e.g., varying illumination and weather conditions). We propose the first deployment-oriented, comprehensive evaluation framework for federated object detection, jointly optimizing model accuracy, computational overhead, memory footprint, and environmental robustness. Extensive empirical studies are conducted on KITTI, BDD100K, and nuScenes using YOLOv5/v8/v11 and Deformable DETR, systematically varying input resolution, batch size, weather conditions, and dynamic client participation. Our results uncover fundamental trade-offs among accuracy, efficiency, and robustness—providing both theoretical insights and actionable engineering guidelines for building trustworthy, scalable federated detection systems in vehicular edge environments.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Object detection is crucial for Connected Autonomous Vehicles (CAVs) to perceive their surroundings and make safe driving decisions. Centralized training of object detection models often achieves promising accuracy, fast convergence, and simplified training process, but it falls short in scalability, adaptability, and privacy-preservation. Federated learning (FL), by contrast, enables collaborative, privacy-preserving, and continuous training across naturally distributed CAV fleets. However, deploying FL in real-world CAVs remains challenging due to the substantial computational demands of training and inference, coupled with highly diverse operating conditions. Practical deployment must address three critical factors: (i) heterogeneity from non-IID data distributions, (ii) constrained onboard computing hardware, and (iii) environmental variability such as lighting and weather, alongside systematic evaluation to ensure reliable performance. This work introduces the first holistic deployment-oriented evaluation of FL-based object detection in CAVs, integrating model performance, system-level resource profiling, and environmental robustness. Using state-of-the-art detectors, YOLOv5, YOLOv8, YOLOv11, and Deformable DETR, evaluated on the KITTI, BDD100K, and nuScenes datasets, we analyze trade-offs between detection accuracy, computational cost, and resource usage under diverse resolutions, batch sizes, weather and lighting conditions, and dynamic client participation, paving the way for robust FL deployment in CAVs.

Problem

• Addressing computational demands for federated learning in CAVs
• Overcoming data heterogeneity and environmental variability challenges
• Evaluating trade-offs between accuracy, cost, and resource usage

Innovation

• Federated learning for privacy-preserving collaborative training
• Deployment-oriented evaluation addressing heterogeneity and constraints
• Integrates model performance, resource profiling, and environmental robustness