Modular Multi-Sensor System for Environment Perception (MOSEP) Dataset Showcase

MOSEP DigiTrans Adverse Weather Dataset This dataset contains multi-sensor measurements recorded by the Modular Hardware and Software System for Multi-Sensor Environment Perception (MOSEP), a ROS 2-based measurement platform designed for environmental monitoring and perception research. The data was collected during a measurement campaign in October 2023 at the DigiTrans vehicle test track in St. Valentin, Austria, using a newly constructed outdoor rain facility approximately 80 m in length. The primary objective of the campaign was to evaluate the performance of different perception sensors under controlled heavy rainfall conditions. The MOSEP setup - equipped with Lidar, radar, and camera - was positioned just outside the rain-affected area, oriented towards an approaching vehicle. The weather stations were positioned inside the rain-affected area, approximately 7 m from the sensor setup. The full dataset comprises **53 ROS 2 bag files covering: - Rain rates: 0, 30, and 100 mm/h - Vehicle speeds: 0–70 km/h from which 3 examplary files are provided here (0, 30 and 100 mm/h at 40 km/h vehicle speed). Note on WS100 data quality: Although rainfall rates were simultaneously recorded using the Lufft WS100 sensor, these data proved unusable due to electromagnetic interference from the steel structure of the rain facility, which disrupted the upward-facing radar-based precipitation measurements. They are nevertheless included for illustrative purposes. Contents Each recording consists of: - ROS 2 bag file (MCAP format) — serialized sensor data recorded with ROS 2 Humble - metadata.yaml — rosbag2 metadata (topics, QoS profiles, message counts, timestamps) - ouster_metadata.txt — Ouster Lidar intrinsic calibration data (beam angles, lidar-to-sensor transform, pixel shift table) Sensor Setup Sensor Model Data Lidar Ouster OS1-64-U02 Raw packets (ouster_msgs/msg/PacketMsg), 1024×10 mode, dual return Radar Smartmicro UMRR-11 Type 132 Point cloud (sensor_msgs/msg/PointCloud2) Camera Raspberry Pi HQ Camera Compressed images (sensor_msgs/msg/CompressedImage), 960×540 @ 10 Hz Weather station (precipitation) Lufft WS100 Meteorological measurements (lufft_wsx_interfaces/msg/LufftWSXXX) — see quality note above Weather station (multi-parameter) Lufft WS501-UMB Meteorological measurements (lufft_wsx_interfaces/msg/LufftWSXXX) ROS 2 Topics Topic Message Type Sensor Description /sensing/lidar/points sensor_msgs/msg/PointCloud2 Ouster OS1-64-U02 Lidar point cloud — strongest return /sensing/lidar/points2 sensor_msgs/msg/PointCloud2 Ouster OS1-64-U02 Lidar point cloud — second strongest return /sensing/lidar/metadata std_msgs/msg/String Ouster OS1-64-U02 Lidar sensor metadata (JSON) /sensing/radar/points sensor_msgs/msg/PointCloud2 Smartmicro UMRR-11 Type 132 Radar cluster detections (position, velocity, RCS) /sensing/cam/image_raw/compressed sensor_msgs/msg/CompressedImage Raspberry Pi HQ Camera Compressed RGB camera images /sensing/aws/ws100_measurements lufft_wsx_interfaces/msg/LufftWSXXX Lufft WS100 Precipitation sensor measurements (unusable — see note) /sensing/aws/ws501_measurements lufft_wsx_interfaces/msg/LufftWSXXX Lufft WS501-UMB Multi-parameter weather station (temperature, humidity, pressure, wind, radiation) /tf_static tf2_msgs/msg/TFMessage — Static sensor-to-sensor transforms /tf tf2_msgs/msg/TFMessage — Dynamic coordinate frame transforms /sensing/lidar/os_replay/transition_event lifecycle_msgs/msg/TransitionEvent Ouster OS1-64-U02 Lidar driver lifecycle state transitions Format & Playback The bag files use the MCAP storage format. To replay, use e.g. Lichtblick. A custom Lichtblick layout that can be imported is provided in Lichtblick_view.json.