This research addresses the need for advanced perception systems in offshore crane operations, driven by the global shift towards sustainable energy and the increasing complexity of offshore wind farms. Offshore crane operations are essential for the installation and maintenance of these structures, and currently rely heavily on the subjective judgement of operators. This dependence contributes significantly to human error, with approximately 60-80% of accidents linked to poor situational awareness and misperception. Huisman Equipment identified the potential for enhancing safety and operational efficiency by employing advanced sensing technologies, such as 3D-Light Detection and Ranging (LiDAR) systems.

A comprehensive review of existing technologies and methodologies for load detection and tracking highlighted the advantages of LiDAR sensors over cameras, radar, and radio-based sensors. Particularly in offshore environments, where visibility is often compromised, LiDAR’s ability to produce high-resolution, three-dimensional point cloud data, coupled with the stability of a fixed-mounted setup, ensures reliable and consistent monitoring of loads during crane operations.

To achieve accurate load detection using a single LiDAR sensor, the study incorporated techniques such as Statistical Outlier Removal (SOR) and voxel grid downsampling for effective data preprocessing. RANSAC and HDBSCAN were employed for robust background removal and clustering, respectively. These methods were seamlessly integrated into the detection architecture, demonstrating reliable performance against key performance indicators (KPIs) and achieving high accuracy under varying conditions.

The research identified suitable motion models for reliably tracking the movement of detected loads, including a constant turn model for horizontal motion and a constant velocity model for vertical motion. These models were integrated with the Unscented Kalman Filter (UKF) for tracking and the Joint Probabilistic Data Association Filter (JPDAF) for data association. Experimental results confirmed the system’s ability to accurately track load positions, maintain precision, and distinguish loads from clutter in dynamic offshore scenarios.

An experimental setup was designed to replicate real-world conditions and collect data to develop and test the proposed LiDAR-based methods. The setup, simulating the operational environment of a vessel-mounted LiDAR system, provided diverse datasets essential for validating the detection and tracking methods. Testing confirmed that the setup effectively replicated offshore conditions, supporting the refinement and evaluation of the system. The developed methods were rigorously evaluated for their accuracy, reliability, and performance in simulated offshore crane operations. The detection system consistently achieved accuracy exceeding 70% in high-performance scenarios, with effective clustering indices surpassing minimum thresholds. Tracking results demonstrated reliable load identification and positional precision, despite minor systematic errors. The methods proved robust and reliable, addressing key challenges in offshore environments.

In conclusion, this study successfully developed and validated a fixed-mounted LiDAR system for load detection and tracking in offshore crane operations. The research provides a practical, technology driven solution to improve safety and efficiency, while future work should explore enhancing the system’s capabilities under adverse weather conditions and integrating additional sensor technologies to further advance situational awareness.