README for the Honeybee Behavior Dataset

Dataset Overview

This dataset supports the analyses and experiments of the paper:

Blaha et al., "Toward Perpetual Occlusion-Aware Observation of Comb States in Living Honeybee Colonies," IROS 2024. DOI: 10.1109/IROS58592.2024.10801380

The dataset provides comprehensive observations of honeybee behaviour within an observation hive captured using an automated robotic system. The primary objective is to facilitate research in precision beekeeping by offering detailed spatial and temporal data on bee activities.

Data Collection Methodology

The data was collected using a robotic system equipped with sensors and cameras designed to non-invasively monitor bee activity within the hive. The system performed systematic scans of predefined locations, capturing images and detecting bee positions and orientations. This approach minimizes disturbance to the bees and ensures consistent data quality. For more details, see Ulrich et al., "Autonomous tracking of honey bee behaviours over long-term periods with cooperating robots," Science Robotics 2024. DOI: 10.1126/scirobotics.adn6848.

Data Structure

The dataset is organized into three main directories:

• train/: Contains individual bee detections with precise timestamps and spatial coordinates used for training of the predictive models.
• test/: Aggregated observations per scan and tile, including the number of detected bees.
• raw_with_overlaps/: Individual bee detections with detailed positional and orientation information used for testing of effective mapping.

Each file is duplicated (xy_0.csv and xy_1.csv), corresponding to the two sides of the observation hive, each monitored by a separate robot.

File Contents

1. train/ Directory

Files in this directory contain individual bee detections with the following columns:

• tile_id: Identifier for the specific location within the hive.
• t: Timestamp of the detection in Unix epoch seconds (UTC).
• x: X-coordinate of the bee's position in meters.
• y: Y-coordinate of the bee's position in meters.

2. test/ Directory

Files here provide aggregated data per scan and tile:

• bag_name: Label of the ROS bag used for recording.
• scan_id: Identifier for the scan session.
• tile_id: Identifier for the specific location within the hive.
• timestamp: Timestamp of the scan in Unix epoch seconds (UTC).
• cam_x: X-coordinate of the camera's position in meters.
• cam_y: Y-coordinate of the camera's position in meters.
• n_bees: Number of bees detected in the scan.

3. raw_with_overlaps/ Directory

This directory contains detailed individual bee detections:

• bag_name: Label of the ROS bag used for recording.
• scan_id: Identifier for the scan session.
• timestamp: Timestamp of the detection in Unix epoch nanoseconds (UTC).
• tile_id: Identifier for the specific location within the hive.
• cam_x: X-coordinate of the camera's position in meters.
• cam_y: Y-coordinate of the camera's position in meters.
• x1: X-coordinate of the bee's bounding box starting position in meters.
• y1: Y-coordinate of the bee's bounding box starting position in meters.
• x2: X-coordinate of the bee's bounding box ending position in meters.
• y2: Y-coordinate of the bee's bounding box ending position in meters.
• angle: Orientation of the bee in radians.

Data Units

• Timestamps: Unix epoch format (UTC). In train/ and test/ directories, timestamps are in seconds; in raw_with_overlaps/, timestamps are in nanoseconds.
• Spatial Coordinates: All positional data (cam_x, cam_y, x, y, x1, y1, x2, y2) are in meters.
• Angles: Orientations (angle) are in radians.

License and Citation

To attribute this dataset in your research, please cite the two corresponding papers:

Blaha et al., "Toward Perpetual Occlusion-Aware Observation of Comb States in Living Honeybee Colonies," IROS 2024. DOI: 10.1109/IROS58592.2024.10801380
Ulrich et al., "Autonomous tracking of honey bee behaviours over long-term periods with cooperating robots," Science Robotics 2024. DOI: 10.1126/scirobotics.adn6848