Published May 25, 2026 | Version v1
Dataset Open

PoC 3 - Synchronized USV Hydrophone Dataset for Ocean Noise Pollution Monitoring (Trials 1 & 2)

Authors/Creators

Description

OpenSwarm PoC 3: Synchronized USV Hydrophone Dataset for Ocean Noise Pollution Monitoring (Trials 1 & 2)

Abstract

This dataset contains acoustic, spatial, and synchronization data collected during Trial 1 (2026.03.04). and Trial 2 (2026.04.20) of the OpenSwarm European Project, specifically supporting Proof of Concept (PoC) 3: Ocean Noise Pollution Monitoring.

The primary objective of this dataset is to provide highly synchronized multi-node acoustic recordings of marine environments to evaluate noise pollution. The setup features a distributed network of Unmanned Surface Vehicles (USVs) configured as sensory buoys, alongside a mobile ground-truth sound source. A critical feature of this dataset is the implementation of a wireless local synchronization system across all nodes, ensuring a common timestamp reference for precise spatiotemporal analysis of the acoustic data.

Description of the Files

  • Trial 1: 2026.03.04
    • 2026_03_04_datasets.zip: dataset
    • 2026_03_04_drone_dji_top_view.mp4: video from the drone camera at the top of the experiment
  • Trial 2: 2026.04.20
    • 2026_04_20_datasets.zip: dataset
    • 2026_04_20_top_view.zip: video from the drone camera at the top of the experiment

Experimental Setup & Hardware

The experimental architecture consists of four primary nodes—three stationary sensory buoys and one mobile ground-truth vessel.

  • Sensory Nodes (x3 USV Buoys):
    • Acoustic Sensors: Each buoy is equipped with an array of three H1a Hydrophones (Aquarian Audio).
    • Acquisition & Compute: Acoustic signals are processed via a HiFiBerry DAC+ ADC HAT connected to a Raspberry Pi.
    • Synchronization: A SmartMesh (LTC5800) wireless synchronization system is deployed across all nodes, providing a common time-span reference updating at a frequency of 2Hz.
  • Ground-Truth Node (x1 Vessel):
    • Vessel Specifications: A 4-meter boat powered by a petrol engine, acting as the primary acoustic source for validation and testing.
    • Spatial Tracking: Equipped with a RTK (Real-Time Kinematic) GPS module for high-precision localization.
    • Synchronization: Integrated with the same SmartMesh (LTC5800) system to perfectly align the boat's spatial data and engine noise events with the acoustic data captured by the sensory buoys.

Data Structure & Formats

The dataset provides raw and processed data streams across multiple standard formats to facilitate robotic framework integration and standalone signal processing.

  • ROS 2 Bags (.db3): Contains the complete serialized message streams for all sensor topics, allowing for full playback of the experimental trials within a Robot Operating System 2 (ROS 2) environment.
  • CSV Files (.csv): Extracted tabular data for straightforward data analysis. These files contain:
    • High-precision RTK GPS positioning (Ground Truth).
    • SmartMesh synchronization timestamps (2Hz).
    • Corresponding ROS time (rostime).
    • Sensor telemetry and metadata.
  • Audio Files (.wav): Uncompressed, raw multi-channel acoustic data captured by the H1a hydrophone arrays, synced to the timeline established by the SmartMesh system.

Topics

/mavros/gpsstatus/gps1/raw
   mavros_msgs/GPSRAW
/rms_stream/data
   msg/StreamData.msg
  • mavros_msgs/GPSRAW
# FCU GPS RAW message for the gps_status plugin
# A merge of mavlink GPS_RAW_INT and 
# mavlink GPS2_RAW messages.

std_msgs/Header header
## GPS_FIX_TYPE enum
uint8 GPS_FIX_TYPE_NO_GPS     = 0    # No GPS connected
uint8 GPS_FIX_TYPE_NO_FIX     = 1    # No position information, GPS is connected
uint8 GPS_FIX_TYPE_2D_FIX     = 2    # 2D position
uint8 GPS_FIX_TYPE_3D_FIX     = 3    # 3D position
uint8 GPS_FIX_TYPE_DGPS       = 4    # DGPS/SBAS aided 3D position
uint8 GPS_FIX_TYPE_RTK_FLOATR = 5    # TK float, 3D position
uint8 GPS_FIX_TYPE_RTK_FIXEDR = 6    # TK Fixed, 3D position
uint8 GPS_FIX_TYPE_STATIC     = 7    # Static fixed, typically used for base stations
uint8 GPS_FIX_TYPE_PPP        = 8    # PPP, 3D position
uint8 fix_type      # [GPS_FIX_TYPE] GPS fix type

int32 lat           # [degE7] Latitude (WGS84, EGM96 ellipsoid)
int32 lon           # [degE7] Longitude (WGS84, EGM96 ellipsoid)
int32 alt           # [mm] Altitude (MSL). Positive for up. Note that virtually all GPS modules provide the MSL altitude in addition to the WGS84 altitude.
uint16 eph          # GPS HDOP horizontal dilution of position (unitless). If unknown, set to: UINT16_MAX
uint16 epv          # GPS VDOP vertical dilution of position (unitless). If unknown, set to: UINT16_MAX
uint16 vel          # [cm/s] GPS ground speed. If unknown, set to: UINT16_MAX
uint16 cog          # [cdeg] Course over ground (NOT heading, but direction of movement), 0.0..359.99 degrees. If unknown, set to: UINT16_MAX
uint8 satellites_visible # Number of satellites visible. If unknown, set to 255

# -*- only available with MAVLink v2.0 and GPS_RAW_INT messages -*-
int32 alt_ellipsoid # [mm] Altitude (above WGS84, EGM96 ellipsoid). Positive for up.
uint32 h_acc        # [mm] Position uncertainty. Positive for up.
uint32 v_acc        # [mm] Altitude uncertainty. Positive for up.
uint32 vel_acc      # [mm] Speed uncertainty. Positive for up.
int32  hdg_acc      # [degE5] Heading / track uncertainty
uint16 yaw          # [cdeg] Yaw in earth frame from north.

# -*- only available with MAVLink v2.0 and GPS2_RAW messages -*-
uint8 dgps_numch    # Number of DGPS satellites
uint32 dgps_age     # [ms] Age of DGPS info
  • msg/StreamData.msg
string timestamp
sensor_msgs/NavSatFix localization
std_msgs/Float64MultiArray rms_data
string node_id
string utc_time
std_msgs/Float64MultiArray raw_data

Files

2026_03_04_datasets.zip

Files (17.3 GB)

Name Size Download all
md5:bace5bae0c8c3f698e67e2e90fff34ba
97.1 MB Preview Download
md5:309a2affbebd5470624ba6909e62c32d
3.7 GB Preview Download
md5:318be1fea519338679a0710b10acf85a
302.6 MB Preview Download
md5:33a9a4c0b59f23f1f3b1f6c5369840fc
13.2 GB Preview Download
md5:47b86c353e4aef3c66b0c17d4c4e08c0
2.5 MB Preview Download

Additional details

Funding

European Commission
OpenSwarm - Orchestration and Programming ENergy-aware and collaborative Swarms With AI-powered Reliable Methods 101093046