Decentralized Federated Learning for Road User Classification in Enhanced V2X Networks

Abstract

Federated Learning (FL) techniques are emerging in the automotive context to support connected automated driving services. Yet, when applied to vehicular use cases, conventional centralized FL policies show some drawbacks in terms of latency and scalability. This paper focuses on decentralized FL solutions, which attempt to overcome such limitations, by introducing a distributed computing architecture: vehicles exchange the parameters of a shared Machine Learning (ML) model via V2V links, without the need of a central orchestrator. Sharing all ML parameters, however, might not be feasible when minimal V2X bandwidth usage is required or the model is highly complex (e.g., extremely deep networks) as in advanced scenarios for high levels of automation. We thus propose a modular decentralized FL solution and we discuss its application to road user classification in a cooperative vehicular sensing use case. The proposed FL solution performs the point cloud processing of Lidar sensor inputs using a PointNet compliant architecture. It enables the exchange of a subset of the model parameters, namely selected ML model layers, optimized for communication efficiency, convergence and accuracy. We use real sensor data extracted from a publicly available dataset to validate the method, focusing on non-uniform scenarios where sensor data are highly unbalanced across the connected vehicles. For all cases, FL is shown to outperform the ego-sensing approach with minimal bandwidth usage.

Deep Learning-Based Cooperative LiDAR Sensing for Improved Vehicle Positioning

“…Furthermore, the developed method does not rely on maps of the driving environment, as in [39], making the cooperative approach potentially applicable to any driving situation, especially when high-definition maps are unavailable. The proposed solution improves upon all our previous works, which provided an initial proof of concept using only single-stage detectors [57], [58], or focused on studying the DA problem alone [37], [38]. Herein, we provide a more complete and robust solution that jointly addresses the DA problem and improves the localization accuracy thanks to the cooperation among vehicles.…”

“…• We extend our frameworks in [37], [38], [57], [58] by integrating both single-stage and double-stage 3D object detectors to provide a comprehensive cooperative localization platform. • We introduce a comparison with state-of-the-art coop-erative SLAM approaches to show the benefits of the proposal in terms of positioning accuracy as well as in reducing the communication overhead.…”

Decentralized federated learning for extended sensing in 6G connected vehicles

“…In this paper, we investigate the FL technology for the classification of road users or objects (here referred to as road actors) based on Lidar data collected by networked vehicles. A preliminary study on FL system for road actor classification has been performed in [42]. Herein, we extend the previous work by proposing a new cooperative solution based on a fully decentralized FL policy, referred to as consensusdriven FL (C-FL).…”