How the Fusion of Onboard Sensors and V2X Data can Improve (or not) the Cooperative Perception of Connected Automated Vehicles

Jul 16, 2026

Upload by Eden

Automated vehicles depend on onboard sensors for environmental perception, yet performance can degrade in adverse weather or when line of sight is obstructed. Cooperative (collective) perception addresses these limits by sharing perception data among Connected and Automated Vehicles (CAVs) via V2X and fusing it with locally sensed information. This study evaluates how sensing measurement errors, V2X packet losses, and GNSS inaccuracies affect the effectiveness of track-to-track fusion. Results show potential improvements in perceived level and range, while highlighting ghost-vehicle challenges that can introduce additional errors.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

Introduction

High-level cooperative perception and fusion process

High-level data fusion

Fusion architecture for onboard and V2X data

How the Fusion of Onboard Sensors and V2X Data can improve (or not) the Cooperative Perception of Connected Automated Vehicles Amir Mohammadisarab, Miguel Sepulcre, Luca Lusvarghi, Javier Gozalvez Uwicore lab., Universidad Miguel Hernández de Elche (UMH), Spain. Email: {amohammadisarab, msepulcre, llusvarghi, [j.gozalvez}@umh.es](j.gozalvez}@umh.es) Abstract— Automated vehicles rely on onboard sensors to perceive their surroundings and navigate autonomously. However, sensor performance may degrade under adverse weather conditions or when line-of-sight is obstructed. Cooperative perception (or collective perception) is expected to mitigate these limitations by enabling Connected and Automated Vehicles (CAVs) to share sensor data and collaboratively enhance situational awareness. Several studies have analyzed the potential of cooperative perception, yet the fusion of V2X data with information from onboard sensors has received limited focus. V2X data may contain errors that affect the quality of the fused data, and hence the effectiveness of cooperative perception. This study analyzes the impact of sensing measurement errors, V2X packet losses, and GNSS inaccuracies on the effectiveness of cooperative perception. The results highlight the potential of cooperative perception to enhance perception levels and range compared to using onboard sensors alone. However, they also identify key challenges related to the generation of ghost vehicles during the fusion process, which must be addressed to prevent V2X data from introducing additional errors when fused with onboard sensor data. Keywords— Cooperative perception, collective perception, V2X, connected automated driving, autonomous vehicles, sensor fusion, track-to-track, association. I. INTRODUCTION Cooperative perception (or collective perception, CP) is essential for improving the perception of connected and automated vehicles (CAVs) [1] . An automated vehicle relying solely on onboard sensors faces two main challenges in perceiving its environment: the inherent limitations of its field of view and sensing range, and the presence of obstructions that can block the sensors’ line-ofsight. Through CP, CAVs can exchange perception data about detected objects (including kinematic and geometric information) via vehicle-to-everything (V2X) communication thereby extending their field of view and sensing range. To do so, data from V2X cooperative perception must be fused with locally acquired onboard sensor data [2] . This process is typically decomposed into three main subprocesses: temporal and spatial alignment, association, and fusion. Due to variations in positioning systems across different vehicles and delays in V2X communication, the received information about detected objects must first be spatially and temporally aligned with onboard sensor data. Next, association algorithms determine which detected objects from different sources (onboard sensors and other CAVs) correspond to the same real-world object. Finally, a fusion algorithm integrates these common observations into a unified representation [2] . The fusion of V2X and onboard data can be performed at various levels, including low, medium, and high levels [3] . Low-level fusion integrates raw sensor data from different sources, while medium-level fusion combines extracted features. High-level fusion operates at the object or track level, where each sensor independently performs filtering and tracking before fusing the outputs from all sensors. While fusing raw data or features provides finer details, it strains the limited available spectrum. Consequently, current ETSI [4] and SAE [5] standards for cooperative perception have opted for a high-level fusion architecture, as CAVs exchange objectlevel data via V2X. Existing research on multi-sensor high-level fusion for autonomous vehicles (e.g. [3],[6]) has demonstrated the importance of effective track-to-track (or object-level) fusion algorithms. However

FAQs

Why is cooperative perception needed for connected automated vehicles?

Relying only on onboard sensors is limited by field of view, sensing range, and obstructions that block line of sight. Cooperative perception enables CAVs to share detected-object information via V2X to extend perception coverage.

What major fusion subprocesses are involved before track-to-track fusion?

The process is decomposed into temporal and spatial alignment, association, and fusion. V2X information must be aligned with onboard sensor data, associated to the same real-world object, and then fused into a unified representation.

Which error sources are analyzed for their impact on V2X and onboard data fusion?

The study analyzes sensing measurement errors, V2X packet losses, and GNSS inaccuracies. These factors affect the quality and effectiveness of the fused data and the resulting cooperative perception.

Jul 16, 2026

Upload by Eden