Metadata Generation Tools for FAIR Mobile Robot Datasets in Forest Environments

Modern mobile robotics research increasingly depends on high-quality datasets that can be shared, reused, and interpreted beyond the context in which they were collected. This is especially true in complex outdoor environments such as forests, where sensing conditions, terrain properties, weather, and vegetation density strongly influence sensor behavior and algorithm performance. Despite the growing demand for open, interoperable datasets, many robotic data collections lack the metadata needed to satisfy the FAIR (Findable, Accessible, Interoperable, Reusable) principles. Building on the recent framework proposed by Motta (2023) in “A Framework for FAIR Robotic Datasets”, there is a need for tools that automatically capture, structure, and package metadata alongside raw sensor data.

The goal of this thesis is to design and implement a metadata-generation toolkit tailored to mobile robot datasets collected in forested environments. The student will survey the FAIR principles, analyze the requirements described in Motta’s framework, and identify which robot- and environment-specific metadata fields are necessary to ensure future usability of the datasets. These metadata will include sensor configurations (e.g., lidar, radar, GNSS/INS, cameras), robot state information, experiment context, geographical descriptors, and environmental conditions (weather, forest type, ground conditions, time of day, seasonal factors). The student will then implement tools that automatically extract, fetch, or compute these metadata fields and store them according to FAIR-compliant formats.

This project will be supervised by Örebro University as part of the ongoing research project RaCOON at the Robot Navigation and Perception lab. Deliverables include a software module for metadata generation, integration with at least one real dataset, a final thesis report, and a demonstration showing metadata creation for a recorded forest experiment.

Project Objectives

• Study the FAIR principles and the methodology proposed in Motta (2023), “A Framework for FAIR Robotic Datasets”.
• Analyze existing ORU’s robot datasets from forest environments and identify essential metadata fields.
• Determine which robot sensors and experiment parameters must be referenced in the metadata, including but not limited to:

­­­­­     º Se­nsor models, calibration files, configuration parameters, coordinate frames, and sampling rates.

      º Robot trajectory, platform parameters, and timestamps.

­     º Environmental descriptors: weather data, vegetation type, terrain features, GPS location, time and season.

• Develop tools that automatically extract or fetch these metadata elements (e.g., from ROS bag files, configuration files, GNSS logs, or public environmental databases).
• Implement export functionality using FAIR-aligned formats (e.g., JSON-LD, ROS 2 metadata conventions, or the structure recommended by Motta’s framework).
• Validate the toolchain on at least one dataset collected in a forested environment and document workflows for future researchers.

Who Should Apply

This project is ideal for students interested in robotics data management, field robotics, and reproducible research.

Required qualifications:

• Good programming skills in Python,
• Experience working with ROS/ROS 2 and handling sensor datasets.

Recommended qualifications:

• Knowledge of geospatial data, environmental data sources, or robot perception systems.