V2AIX: A Multi-Modal Real-World Dataset of ETSI ITS V2X Messages in Public Road Traffic

TL;DR

This work introduces V2AIX, the first multi-modal real-world dataset of ETSI ITS V2X messages collected in public road traffic, addressing the scarcity of large-scale ETSI ITS data. It documents a measurement campaign yielding over 285k ETSI ITS messages across CAM, DENM, MAPEM, and SPATEM, enriched with multi-sensor context data and precise localization. The authors also present etsi_its_messages, an open-source ROS/ROS 2 toolkit that automatically generates ROS messages from ASN.1 ETSI ITS definitions and provides a bridge to ROS for bidirectional V2X integration. Together, these contributions enable researchers to analyze real-world V2X usage, evaluate localization accuracy, and prototype ROS-based automated driving applications with standardized ETSI ITS messages, paving the way for future expansion and cooperative perception research.

Abstract

Connectivity is a main driver for the ongoing megatrend of automated mobility: future Cooperative Intelligent Transport Systems (C-ITS) will connect road vehicles, traffic signals, roadside infrastructure, and even vulnerable road users, sharing data and compute for safer, more efficient, and more comfortable mobility. In terms of communication technology for realizing such vehicle-to-everything (V2X) communication, the WLAN-based peer-to-peer approach (IEEE 802.11p, ITS-G5 in Europe) competes with C-V2X based on cellular technologies (4G and beyond). Irrespective of the underlying communication standard, common message interfaces are crucial for a common understanding between vehicles, especially from different manufacturers. Targeting this issue, the European Telecommunications Standards Institute (ETSI) has been standardizing V2X message formats such as the Cooperative Awareness Message (CAM). In this work, we present V2AIX, a multi-modal real-world dataset of ETSI ITS messages gathered in public road traffic, the first of its kind. Collected in measurement drives and with stationary infrastructure, we have recorded more than 285 000 V2X messages from more than 2380 vehicles and roadside units in public road traffic. Alongside a first analysis of the dataset, we present a way of integrating ETSI ITS V2X messages into the Robot Operating System (ROS). This enables researchers to not only thoroughly analyze real-world V2X data, but to also study and implement standardized V2X messages in ROS-based automated driving applications. The full dataset is publicly available for non-commercial use at v2aix.ika.rwth-aachen.de.

Figures (5)

• Figure 1: V2X messages at core recording locations collected in the V2AIX dataset: driven routes in blue, reference positions of received V2X messages colored by unique ITS station. From top-left to bottom-right: overview, A44, Dusseldorf, Aachen (Aix-la-Chapelle) and Cologne.
• Figure 2: Concept of etsi_its_messages: 1) Code Generation: based on ASN.1 definitions, C++ libraries containing C-struct implementations of ETSI ITS message types including encoding/decoding support are generated using asn1cLevWalkin_asn1c (coding); ROS message type definitions are generated (msgs); conversion functions between C-structs and ROS message objects are generated (conversion); 2) Runtime: V2X messages are received/transmitted with dedicated hardware, e.g., via ITS-G5, or simulated; a simple driver publishes UDP message payloads as ROS messages; the etsi_its_conversion node decodes the payloads and converts them to the ROS messsage equivalents of ETSI ITS message types; downstream ROS applications like RViz for visualization work with ROS V2X messages.
• Figure 3: Automated data collection procedure: $t_0$) GNSS data is recorded to a new ROS bag file; $t_1$) no ETSI ITS V2X message has been received for $\Delta t_A = t_1 - t_0$, recording to a new ROS bag file is started; $t_2$) ETSI ITS V2X messages are received and recorded; $t_3$) reference position of received CAM is within $d_{\mathrm{min}}$, camera images and lidar point clouds are recorded; $t_4$) received CAM is beyond $d_{\mathrm{min}}$; $t_5$) no received CAM has been within $d_{\mathrm{min}}$ for $\Delta t_B = t_5 - t_4$, recording of camera and lidar data is stopped; $t_6$) last ETSI ITS V2X message for a while is received; $t_7$) no ETSI ITS V2X message has been received for $\Delta t_A = t_7 - t_6$, recording to a new ROS bag file is started. A single bag file may contain multiple sequences of V2X messages and sensor data.
• Figure 4: Number of unique vehicles per object dimension identified in all CAMs
• Figure 5: CAM trajectories collected by an urban (left) and a highway (right) stationary infrastructure station