An issue that ROVs experience during operations is disturbances from the tether, making navigation and control more difficult as real-time measurements are not currently available. This paper proposes the development of an innovative sensor that can measure tether forces in multiple degrees of freedom. These tether forces apply an external disturbance during operation, which is difficult to model and predict. The sensor provides real-time input on the effect the tether has on the ROV, which can be utilized in feed-forward in the control system in combination with a feedback loop. There are 2 proposed designs: a 4 DOF sensor design using a plastic bottle and a 6 DOF version utilizing an aluminum cross with hollowed sections. Both designs use strain gauges to measure and determine the direction and magnitude of the force from the tether.
The sensors are implemented to a modified BlueROV2 using ROS. Station-keeping tests in a harbor and test basin are done for the 4 DOF version to evaluate performance. The sensor shows potential, improving response in heave but worsening it in yaw. It removes and adds oscillations both in frequency and amplitude depending on the orientation of the waves relative to the sensor. Indicating alternative control strategies might be more suitable. The 6 DOF version is not tested on the BlueROV2. In future work, additional development is required to ensure the viability of the tether force sensor as a commercial product.
Continuous inspection and mapping of the seabed allows for monitoring the impact of anthropogenic activities on benthic ecosystems. Compared to traditional manual assessment methods which are impractical at scale, computer vision holds great potential for widespread and long-term monitoring.
We deploy an underwater remotely operated vehicle (ROV) in Jammer Bay, a heavily fished area in the Greater North Sea, and capture videos of the seabed for habitat classification. The collected JAMBO dataset is inherently ambiguous: water in the bay is typically turbid which degrades visibility and makes habitats more difficult to identify. To capture the uncertainties involved in manual visual inspection, we employ multiple annotators to classify the same set of images and analyze time spent per annotation, the extent to which annotators agree, and more.
We then evaluate the potential of vision foundation models (DINO, OpenCLIP, BioCLIP) for automating image-based benthic habitat classification. We find that despite ambiguity in the dataset, a well chosen pre-trained feature extractor with linear probing can match the performance of manual annotators when evaluated in known locations. However, generalization across time and place is an important challenge.
A crucial component for unmanned underwater vehicles (UUVs), including remotely operated vehicles (ROVs) and autonomous underwater vehicles (AUVs), are the thrusters, which, in addition, are sensitive to damage during operations in harsh environments. This paper presents a study on the impact of incipient faults on the performance of thruster propellers used in offshore operations. The study evaluates the reduction in propeller performance due to wear and tear under realistic working conditions. The study employs a combination of experimental data analysis and signal processing techniques, including fast Fourier transforms and harmonics analysis, to identify faults and assess their severity. The results show that worn propellers can be identified through 5th-order harmonics and rotational velocity changes. The paper concludes with a proposal for future research using a model-based approach to enhance fault detection capabilities further.
Monitoring methods, such as seabed bottom-towed cameras, sediment grabs, and benthic sledges, have limitations in spatial coverage, cause seabed disturbance, are restricted to soft-bottom substrates, and offer low flexibility for marine seabed monitoring. In this study, we investigate the potential of a non-invasive and simple underwater remotely operated vehicle (ROV) to enhance marine seabed monitoring. A tethered ROV equipped with a GoPro camera was deployed in three areas of Skagerrak at depths from 15-34 m to assess accuracy in species identification and substrate classification identified from still frames. The quality of still frames varied between areas due to turbidity, motion blur, and marine snow, which reduced the number of high-quality frames by approximately 20%. Classification of substrates and taxa identification were possible in the remaining still frames. Two different substrates were detected: sand and stone reef. Stone reefs had a lower occurrence compared to sand. A total of 10 taxa were detected in the two substrate types. The highest abundance was observed in the stone reef substrate compared to the sand substrate. Identification at the species level was limited due to the quality of the still frames, which affected the detectability of morphological traits. This study demonstrates that a widely accessible ROV can be used for marine monitoring. The ROV can be used in different substrates, and still frames provide valuable information on species composition, which can enhance the replicability of monitoring programs.