Maritime Vessel Tank Inspection using Aerial Robots: Experience from the field and dataset release

TL;DR

The paper tackles the hazardous and costly problem of inspecting maritime ballast tanks by delivering an autonomous aerial robot system (RMF-Owl) with onboard SLAM and graph-based exploration tailored to multi-compartment ballast tanks. It demonstrates field deployments across three vessels and seven tank sections, detailing hardware, software, and autonomous navigation strategies, including inter-compartment passage through manholes and multi-modal sensing. Key contributions include the onboard CompSLAM-based localization, GBPlanner-driven exploration/inspection, and a public ballast_water_tank_dataset that accompanies the work. The findings highlight resilient autonomy, semantic-aware inspection planning, and the critical role of lighting for defect detection, offering a pathway to safer, more efficient maritime tank inspections and a valuable dataset for the research community.

Abstract

This paper presents field results and lessons learned from the deployment of aerial robots inside ship ballast tanks. Vessel tanks including ballast tanks and cargo holds present dark, dusty environments having simultaneously very narrow openings and wide open spaces that create several challenges for autonomous navigation and inspection operations. We present a system for vessel tank inspection using an aerial robot along with its autonomy modules. We show the results of autonomous exploration and visual inspection in 3 ships spanning across 7 distinct types of sections of the ballast tanks. Additionally, we comment on the lessons learned from the field and possible directions for future work. Finally, we release a dataset consisting of the data from these missions along with data collected with a handheld sensor stick.

Figures (5)

• Figure 1: Instance of RMF-Owl autonomously operating in a double bottom section of a ballast tank inside a Floating Production Storage and Offloading (FPSO) vessel.
• Figure 2: Illustration of the sensing and computing on the RMF-Owl aerial robot and Mjolnir handheld sensor stick. All the autonomy modules including SLAM, Path planning, and Control run completely onboard the robot.
• Figure 3: Final maps, as generated onboard the robot, along with instances of the RMF-Owl aerial robot performing exploration and inspection mission and data collection using Mjolnir in FPSO1. We conducted a total of $3$ autonomous and $1$ manual flights using RMF-Owl and $4$ handheld data collection missions using Mjolnir.
• Figure 4: Final maps, as generated onboard the robot, along with instances of deployment of the RMF-Owl aerial robot in FPSO2. The robot was deployed in the side sections (1),(3), double bottom (d.b.) section (2.1), and bilge (2.2) section. We present both autonomous exploration-inspection missions and manual flights. The dataset covers missions in vastly different sections, navigation through extremely narrow manholes ($0.7m \times 0.5m$), and multi-level missions.
• Figure 5: Final maps, as generated onboard the robot, along with instances of the RMF-Owl aerial robot performing exploration and inspection missions in the Oil Tanker. All $5$ tests were conducted in the side tanks covering $3$ - $6$ compartments. The environment contains two types of manholes $0.8m \times 0.6m$ and $1.2m \times 0.6m$.