December 2020 OES Beacon Events

The Student Poster Competition at Global OCEANS 2020

Shyam Madhusudhana, OES Student Poster Competition Chair

The Student Poster Competition (SPC) is a flagship event of the MTS/OES OCEANS conferences in which undergraduate and graduate students from colleges and universities around the world participate. The SPC was envisioned and created by Col. Norman Miller and was first implemented in 1989. It has been a feature of OCEANS conferences ever since. Typically, 15-20 students are selected to participate in the Competition, based on reviews (two stages) of their abstracts. The selected students get an opportunity to present, at the OCEANS conference, a poster describing their work.

Given the pandemic-induced lockdowns and travel restrictions around the globe, this year’s Singapore and Gulf Coast OCEANS conferences were merged into a unified virtual event, Global OCEANS 2020, in ensuring continuity in the OCEANS conferences series. The SPCs of the constituent editions of the conference, however, were held independently, with 11 and 13 student participants in the Singapore and Gulf Coast editions, respectively. For this SPC, the participating students were required to switch to electronic posters. Poster formats were prescribed beforehand to ensure that the submitted posters were conducive for online presentations. The students were also required to submit a pre-recorded 5-minute video of their presentations providing a quick overview of their study orally. Interactions between participants and judges were facilitated via managed one-on-one Q&A sessions during the conference.

I would like to congratulate the Local Organizing Committee (LOC) General Chairs and the LOC SPC Chairs for beating all pandemic-related predicaments and “pulling off” a successful SPC event. The SPC leadership included LOC SPC Chairs Too Yuen Min (from Singapore OCEANS) and Stephan Howden (from Gulf Coast OCEANS). I also take this opportunity to thank the sponsors—Office of Naval Research, Office of Naval Research-Global, Schmidt Ocean Institute and National Oceanic and Atmospheric Administration—for their support of this year’s SPC and also for their continued support for OCEANS conferences in general.

The full list of participants (including the prize winners) together with their affiliation, poster title and an abstract of their poster are given below.

Singapore SPC

First Prize, Norman Miller Award (Certificate & $3000 prize)

Marija Jegorova, The University of Edinburgh, UK

Unlimited Resolution Image Generation with R2D2-GANs

Abstract—In this paper we present a novel simulation technique for generating high quality images of any predefined resolution. This method can be used to synthesize sonar scans of size equivalent to those collected during a full-length mission, with across track resolutions of any chosen magnitude. In essence, our model extends Generative Adversarial Networks (GANs) based architecture into a conditional recursive setting, that facilitates the continuity of the generated images. The data produced is continuous, realistically-looking, and can also be generated at least two times faster than the real speed of acquisition for the sonars with higher resolutions, such as EdgeTech. The seabed topography can be fully controlled by the user. The visual assessment tests demonstrate that humans cannot distinguish the simulated images from real. Moreover, experimental results suggest that in the absence of real data the autonomous recognition systems can benefit greatly from training with the synthetic data, produced by the R2D2-GANs.

Second Prize (Certificate and $2000 prize)

Matias Carandell, Universitat Politecnica de Catalunya, Spain

Impact on the Wave Parameters Estimation of a Kinetic Energy Harvester Embedded into a Drifter

Abstract—The effect of a Kinetic Energy Harvester (KEH) on the wave parameters estimation at a WAVY Ocean (WO) drifter is being studied. An algorithm has been developed to calculate the wave parameters from the Inertial Measuring Unit (IMU) embedded on the drifter. Simulations performed by OrcaFlex have been used to refine the algorithm and assess the measurement errors derived from the drifter response. Finally, a WO prototype has been deployed in the controlled environment of CIEM wave flume. Results prove that the KEH has no significant impact on the wave parameter estimation.

Third Prize (Certificate and $1000 prize)

Eduardo Ochoa Melendez, University of Girona, Spain

Allowing untrained scientists to safely pilot ROVs : Early collision detection and avoidance using omnidirectional vision

Abstract—The study of underwater environments involves multiple hazards that can compromise the safety of robots. Underwater missions require a high level of attention from Remotely Operated Vehicle (ROV) operators to avoid damage to the robot. For this reason, there is a growing trend in research to develop systems with new capabilities, such as advanced assisted mapping, spatial awareness and safety, and user immersion.

The aim of this work is to devise a system that provides the vehicle with proximity awareness capabilities for navigation in complex environments. By using the advantages of omnidirectional multi-camera systems, a much higher level of spatial awareness can be achieved. This paper presents a visual-based multi-camera system which is able to detect the presence of nearby objects in the environment, to create a local map of points, and to assign collision risk values to this map. The system exploits this information to generate warnings when approaching potentially dangerous obstacles and at the same time creates a collision risk map that provides a proximity awareness representation of the environment.

Matteo Bresciani, Università di Pisa, Italy

Dynamic parameters identification for a longitudinal model of an AUV exploiting experimental data

Abstract—This work addresses the problem of determining a low cost method to identify the dynamic parameters of the surge motion model for an Autonomous Underwater Vehicle (AUV) utilising experimental data. With this goal, various sea tests were planned and carried out, involving minimal equipment. In particular, an acoustic tracking system composed of passive Direction of Arrival (DoA) sensors was used to track the vehicle while moving underwater. Such dataset was exploited to identify the values of the dynamic parameters with a best fit approach. Eventually, the comparison between the signals computed through the identified model and the experimental data has been reported as a preliminary evaluation.

Hui-Min Chou, National Sun Yat-sen University, Taiwan

Development of a Monocular Vision Deep Learning-based AUV Diver-Following Control System

Abstract—Divers have a variety of tasks typically with high complexity and high risk. Autonomous underwater vehicles (AUVs) can improve the efficiency and safety of divers by assisting divers in performing underwater operations. A dive-buddy AUV should have the ability to follow a diver and interact with a diver. Dive-buddy AUVs can be developed based on machine vision technology or acoustics technology, leading to different advantages and limitations regarding the diver-following and human-machine interaction abilities of the AUVs. Compared with sonar devices, the main limitation of optical cameras is their short underwater visibility range. However, optical cameras have the advantages of high resolution, high frame rate, low cost, and high application popularity. The IUT AUV, developed by the Institute of Undersea Technology at Nation Sun Yat-sen University, is the testbed AUV of this research. This research aimed to make the IUT AUV become a diver-following AUV. This research applied Tiny-YOLOv3 convolutional neural network (CNN) to image detection of divers and developed a diver detection module as a payload sensor for the IUT AUV. This research developed a single-diver following control algorithm and evaluated the single-diver following performance under different scenarios through hardware-in-the-loop simulation (HIL) system that conforms to the communication format and power architecture of the IUT AUV. This research verified the single-diver following capability of the IUT AUV through closed water experiments conducted in a towing tank.

Basil Hable, University of Wisconsin-Milwaukee, USA

A novel particle sensor for the detection of pollution intrusion in harbors
Abstract—The University of Wisconsin – Milwaukee’s School of Freshwater Sciences is built at a port serving building materials companies, a grain elevator complex, a salt warehouse, a nearby oil company, an industrial refuse dumping ground, and has the city of Milwaukee, Wisconsin, USA’s sewage treatment plant across the river. Naturally, this area is a testbed for finding environmentally-sound business solutions for public and private entities at the confluence of the Kinnickinnic River and Milwaukee Municipal Mooring Basin. Our group has developed an optical sensor for detecting particulate matter suspended in water at a lower cost and faster speeds than traditional methods. This manuscript will discuss the implementation and effectiveness of the sensor for detecting changes in the port’s water quality.

Mingi Jeong, Dartmouth College, USA

Catabot: Autonomous Surface Vehicle with an Optimized Design for Environmental Monitoring

Abstract—This paper presents an optimized design of research-oriented ASVs and a systematic design evaluation methodology for reliable in-water sensing. The objective is to minimize the interference on sensor readings by any ASV maneuver. The design space includes motors and sensors locations. In addition, this paper analyzes modularity – i.e., the effects of new sensor’s installation. All prototype designs are thoroughly tested using hydrostatic analyses, Computational Fluid Dynamics (CFD) simulations, and real-world field testings. Quantitative metrics, including trim, pitch, velocity magnitude of flow, and turbulence, are used to compare different configurations. Our experiments show that a motor configuration at the back part of the straights hulls is the most optimal design, resulting in high-quality data collection.

Nare Karapetyan, University of South Carolina, USA

Coverage Path Planning for Mapping of Underwater Structures

Abstract—This paper addresses the problem of the coverage path planning in a 3D environment for surveying underwater structures. We propose to use the navigation strategy that a human diver will execute when circumnavigating around a region of interest, in particular when collecting data from a shipwreck. In contrast to the previous methods in the literature, we are aiming to perform coverage in completely unknown environment with some initial prior information. Our proposed method uses convolutional neural networks to learn the control commands based on the visual input. Preliminary results and a detailed overview of the proposed method are discussed.

Byeongjin Kim, Pohang University of Science and Technology, Republic of Korea

Imaging Sonar based AUV Localization and 3D Mapping using Image Sequences

Abstract—This paper proposes a three-dimensional (3D) sonar mapping method and autonomous underwater vehicle (AUV) localization method using two-dimensional (2D) sonar image sequences of an imaging sonar. The 2D sonar image sequences are registered to generate a 2D mosaic sonar map. In this process, we can estimate the displacement and rotation relationship between the sonar image pairs, and use this to estimate the position of the AUV. A 3D point cloud is generated from 2D sonar image sequences. This method takes advantage of the mobility of the AUV to reconstruct the height information, and partially solves the ambiguity issues in the elevation angle of the imaging sonar. The height map is generated by accumulating the 3D point cloud. By fusing two maps, we can generate a 3D sonar map. We verified the proposed method by conducting field experiments. Three types of artificial reef on the seabed were scanned, and a 3D sonar map was created. In the process, the trajectory of the AUV was estimated and compared with the trajectory of the dead reckoning.

Kexin Li, National University of Singapore, Singapore

Informative Path Planning for Acoustic Source Localization With Environmental Uncertainties

Abstract—Source localization in the context of underwater environment has been recognized as an important but challenging research topic. Conventional methods usually require a receiver array with accurate time synchronization and a large number of elements. The feasibility of accurately localizing an underwater source by a low-cost and small underwater vehicle is greatly limited. Our previous work has successfully applied matched field processing concept to do single-vehicle underwater source localization with informative path planning. However, it requires accurate environmental knowledge of entire search space. In this paper, we focus on improving the robustness of our originally proposed approach. This improvement aims to ensure that our proposed source localization approach performs well even with the presence of environmental uncertainties. Simulation studies show that our robust method is capable of mitigating the effect of environmental mismatch in the MFP model for source localization.

XiaoGang Li, Ocean University of China, China

Extended State Observer Based Iteration Learning Fault-tolerant Control Scheme for AUV

Abstract—To deal with the fault of the thrusters of AUV, an iterative learning algorithm based on the extended states observer (ESO) is established. In this algorithm, the nonlinear feedback mechanism of the ESO is transplanted to iterative learning processes, that is, the nonlinear function of the current output residual is used to adjust the value of the virtual fault in the next iteration. The obtained result shows the favorable convergence speed and the fault estimation precision of the proposed method.

Gulf Coast OCEANS

First Prize, Norman Miller Award (Certificate & $3000 prize)

Jordan Pierce, University of New Hampshire, USA

Reducing Annotation Times: Semantic Segmentation of Coral Reef Survey Images

Abstract—Benthic quadrat studies requiring time-intensive manual image annotation are currently a critical component of assessing the health of coral reefs. Patch-based image classification using convolutional neural networks (CNNs) can automate this task by providing sparse labels, but remain computationally inefficient. This work extends the idea of automatic image annotation by using fully convolutional networks (FCNs) to provide dense labels through semantic segmentation. We present an improved version the Multilevel Superpixel Segmentation (MSS) algorithm, which repurposes existing sparse labels for images by converting them into the dense labels necessary for training a FCN automatically. Our implementation—Fast-MSS—is demonstrated to perform considerably faster than the original without sacrificing accuracy. To showcase the applicability to benthic ecologists, we validate this method using the Moorea Labeled Coral (MLC) dataset as a benchmark. FCNs are evaluated by comparing their predictions on test images with the corresponding ground-truth sparse labels. Our results indicate that FCNs’ perform with accuracies that are suitable for many ecological applications, and can increase even further when trained on dense labels augmented with additional sparse labels provided by a patch-based image classifier. The contributions of this study help move the field of benthic ecology towards more efficient monitoring of coral reefs through entirely automated processes.

Second Prize (Certificate and $2000 prize)

Francesco Ruscio, University of Pisa, Italy

Geo-referenced visual data for Posidonia Oceanica coverage using audio for data synchronization

Abstract—The supervision of underwater areas is essential for the preservation of marine ecosystems. In the Italian context, Professional divers of Environmental Protection Agencies are periodically involved in monitoring activities that are repetitive, dangerous, and expensive. The Ligurian Regional Agency for the Environmental Protection (ARPAL) and the University of Pisa (UNIPI) are collaborating towards the employment of robots and ICT tools to improve the monitoring activities in terms of safety, cost and time effectiveness. This paper reports a strategy to geo-reference underwater visual data using audio for data synchronization. The work refers to a visual dataset acquired by a Smart Dive Scooter (SDS) during a Posidonia Oceanica (Po) monitoring activity in front of the Ligurian Coast. The proposed strategy concerns the synchronization between the audio track recorded by a camera and the transponder pings adopted for the SDS acoustic positioning system. The paper also reports the exploitation of the geo-referenced optical data for the identification of Po meadow over the mission area using a Machine Learning algorithm. The results are very promising and can lead to an accurate geo-referenced identification of Po and the reconstruction of the surveyed area.

Third Prize (Certificate and $1000 prize)

Kavita Varma, Florida Atlantic University, USA

Autonomous Plankton Classification from Reconstructed Holographic Imagery by L1-PCA-assisted Convolutional Neural Networks

Abstract—Monitoring and characterization of plankton species abundance and distribution in both freshwater and marine environments around the world is key for forecasting harmful algal bloom events that can cause substantial damage to aquatic ecosystems as well as the local economy. In this paper, we leverage reconstructed imagery from a novel submersible holographic system to test experimentally a new data analysis tool for robust training of a convolutional neural network (CNN) toward rapid high-confidence classification of plankton. Manual labeling oft he entire plankton dataset with the appropriate class is often prohibitively costly, therefore the dataset may contain mislabeled instances. We test a new method based on L1-norm principal-component analysis to improve the quality of labeled data sets. We carry out experiments with four classes of plankton that may contain images with wrong labels. First, the training data in each class is contaminated with images that belong to one of the four classes. Then, we consider images from other plankton species and images of background scenery that contaminate the training data of the four classes. We show that L1-norm tensor-conformity curation of the data identifies and removes inappropriate class instances from the training set and drastically improves the classification accuracy of a miniVGG CNN in both cases.

Xinwei Chen, Memorial University of Newfoundland, Canada

A Method to Mitigate the Influence of Rain on Wind Direction Estimation From X-band Marine Radar Images

Abstract—In this paper, a novel scheme is employed to mitigate the influence of rain on wind direction estimation from X-band marine radar data. Rain-contaminated regions with blurry wave signatures are first identified using the self-organizing-map(SOM)-based clustering model. Pixels located in those regions are then corrected using a deep-neural-network-based image dehazing model called DehazeNet. After obtaining the rain-corrected radar image, wind direction can be obtained using the classic curve-fitting-based method. Shipborne marine radar data collected under precipitation in a sea trial off the east coast of Canada are utilized to validate the proposed scheme. Experimental results show that the accuracy of wind direction estimation is significantly improved using the proposed scheme with a reduction of root mean square deviation (RMSD) by 19.1◦.

Mohammad Imran Chowdhury, Florida State University, USA

The PRM-A* Path Planning Algorithm For UUVs: An Application To Navy Mission Planning

Abstract—This paper presents a novel path planning algorithm for unmanned underwater vehicles (UUVs) that arises by combining two well-known algorithms in such a way as to exploit their respective advantages while mitigating their deficiencies. The two algorithms are A-star (A*) and Probabilistic Road Map (PRM). The operational environment is given as a portion of open sea interspersed with various obstacles. The free space is the portion not occupied by obstacles.

A* employs heuristic search to explore the free space for a path from the start to the goal. This algorithm always returns the shortest (optimal cost) path. A drawback is that the path sometimes runs dangerously close to an obstacle. In addition, the computed path can have sharp turns that are difficult for the UUV to negotiate.

PRM starts with a random sampling of the points in the free space. Once some predetermined number of such points are chosen, each point is paired with some of it’s free space neighbors, and these pairs are recorded in what is called a road map. Then Dijkstra’s algorithm is applied to the road map using nearest neighbor information to find a path from the start node to the goal. This has the advantage that the resulting path is smooth (no sharp turns) and stays safely away from the obstacles. It has the drawback that the path is not optimal; it can be much longer than necessary.

The idea presented in this paper is to first run both A* and PRM on the given operational environment. Then take the region bounded by these two lines as a new (smaller) operational environment, and rerun PRM. This produces a new path that more closely approximates the A* path while at the same time is smooth and, with proper monitoring, is assured to keep a safe distance from the obstacles. This procedure can be repeated as many times as desired, each time producing a better approximation of the optimal A* path.

While the focus here has been on UUVs, the same approach can be applied for unmanned surface vehicles (USVs).

Johnson Oguntuase, University of Southern Mississippi, USA

Vertical Accuracies of Mass-Market GNSS Receivers and Antennas in Ellipsoid Reference Survey Strategy for Marine Applications

Abstract—This paper describes the vertical accuracies achievable using Low-power, Mass-market, Multifrequency, Multi-constellation GNSS (LM3GNSS) receivers, and antennas, and evaluates the results as a precursor to providing alternatives to high-end GNSS hardware on emerging remotely operated and unmanned systems, such as autonomous surface vehicles (ASV), small unmanned aircraft systems (sUAS), and offshore GNSS buoys. The LM3GNSS receivers are relatively new in the market, and they are affordable ($243 – $2,670). Given their low-power characteristics, they readily fulfill the power budget requirement aboard marine systems. In the first round of our studies, we conducted five experiments in a minivan, which traveled at approximately 80 km per hour (50 mph) with matching pairs of receivers and antennas. We assume that the results should be comparable to solutions aboard a marine vessel at typical hydrographic survey speeds (10 – 20 km per hour {5 – 10 knots}). The matching pairs of receivers include four LM3GNSS hardware, Trimble NetR9 receiver, and Zephyr3 antenna. The results show that LM3GNSS receivers are capable of achieving vertical positioning uncertainties that are within 10 cm of the geodetic results (95% confidence level).

Özer Özkahraman, KTH Royal Institude of Technology, Sweden

Underwater Caging and Capture for Autonomous Underwater Vehicles

Abstract—In this paper, we consider the problem of caging and eventual capture of an underwater entity using multiple Autonomous Underwater Vehicles (AUVs) in a 3D water volume We solve this problem both with and without taking bathymetry into account. Our proposed algorithm for range-limited sensing in 3D environments captures a finite-speed entity based on sparse and irregular observations. After an isolated initial sighting of the entity, the uncertainty of its whereabouts grows while deployment of the AUV system is underway. To contain the entity, an initial cage, or barrier of sensing footprints, is created around the initial sighting, using islands and other terrain as part of the cage if available. After the initial cage is established, the system waits for a second sighting, and the possible opportunity to create a smaller, shrinkable cage. This process continues until at some point it is possible to create this smaller cage, resulting in capture, meaning the entity is sensed directly and continuously. We present a set of algorithms for addressing the scenario above, and illustrate their performance on a set of examples. The proposed algorithm is a combination of solutions to the min-cut problem, the set cover problem, the linear bottleneck assignment problem and the Thomson problem.

Amy Phung, Olin College of Engineering, USA

A comparison of Biogeochemical Argo sensors, remote sensing systems, and shipborne field fluorometers to measure Chlorophyll-A concentrations in the Pacific Ocean off the northern coast of New Zealand

Abstract—Accurately measuring chlorophyll a concentrations within the world’s oceans is an important part of building our understanding of its underlying processes and the human impact on it, and developing tools to do this is an area of active study. Some methods used today to collect this data include in-situ fluorometers on board automated Biogeochemical Argofloats, flow-through fluorometers on board ocean going vessels, and ocean color algorithms applied to remote sensing data. While shipborne field fluorometers are the most accurate of the three since they can be recalibrated before and after each expedition, they are limited in spatial and temporal coverage due to their dependence on expensive oceanographic research cruises. The Biogeochemical floats help to increase the coverage of fluorometer data by automating the data collection, but are known to suffer from sensor drift over time since their fluorometers cannot be serviced and calibrated regularly. Remote sensing data has by far the greatest spatial and temporal coverage of the three methods, but is known to be significantly less accurate in certain regions and is limited to surface measurements. This study compares these three measurement methods by analyzing data collected by a 10AU Field and Laboratory Fluorometer connected to a flow-through system, data from a Biogeochemical Argo float, and satellite data from the VIIRS-SNPP dataset in the same region. The results of comparisons between each of these collection methods are presented.

Maria Pereira, University of Porto, Portugal

Detecting Docking-based Structures for Persistent ASVs using a Volumetric Neural Network

Abstract—In recent years, research concerning the operation of Autonomous Surface Vehicles (ASVs) has seen an upward trend, although the full-scale application of this type of vehicles still encounters diverse limitations. In particular, the docking and undocking processes of an ASV are tasks that currently require human intervention. Aiming to take one step further towards enabling a vessel to dock autonomously, this article presents a Deep Learning approach to detect a docking structure in the environment surrounding the vessel. The work also included the acquisition of a dataset composed of LiDAR scans and RGB images, along with IMU and GPS information, obtained in simulation. The developed network achieved an accuracy of 95.99%, being robust to several degrees of Gaussian noise, with an average accuracy of 93.34% and a deviation of 5.46% for the worst case.

Eivind Salvesen, NTNU, Norway

Robust methods of unsupervised clustering to discover new planktonic species in-situ

Abstract—Plankton species are of vital importance to the marine food chain. They are susceptible to minor changes in their environment, which can lead to rapid and devastating changes in the global ecosystem. Thus, monitoring plankton species and their population dispersion is crucial to understanding the dynamics of their community abundance as well as their consumers in higher trophic levels. Technological advancements of systems providing high-resolution imaging augmented by powerful computing devices made it possible to infer the distribution from sampling millions of planktonic images at low cost. Yet, this requires an extensive and time consuming manual labeling effort. The process of training to distinguish different species on manually labeled data is called supervised learning. The objective of this paper is to find new algorithms capable of minimizing the training supervision and assisting in discovering unseen classes. We explore the use of unsupervised classes of models for in-situ classification and identification of planktonic images. The aim is to embed those models into existing robotic imaging platforms to enhance the classification ability and to allow the discovery of new classes without any prior knowledge or exhaustive labeling effort. This work compares different models and shows their abilities to learn essential data structures over the National Science Bowl planktonic dataset.

Lumi Schildkraut, University of Rochester, USA

Investigating the Role of Spatiotemporal Optical Beam Profiles in Mixed Layer Oceanic Communication Channels

Abstract—Underwater optical communication channels can provide access to GHz speed communications between assets located within a few hundred meters range of each other. However, channel phenomena such as turbulence, scattering layers, and propagation through density gradients (pycnoclines) even in the clearest of natural waters, can significantly degrade channel performance. Here, the role of the spatiotemporal structure of the optical resource is considered as a mechanism for improving performance of a surface-to-subsea downlink scenario through a typical pycnocline in clear, oceanic waters. A 3D+1 numerical solver was developed to incorporate experimentally collected CTD measurements of a pycnocline in the upper 80 m of the mixed layer in order to evaluate and compare the propagation of semi non-diffracting Bessel-Gauss beams and standard Gaussian beams through Jerlov Case 1 waters. In certain link environments with certain initial beam parameters, we find that utilizing nonstandard optical resources can provide system performance advantages.

Andreas Teigen, NTNU, Norway

Leveraging Similarity Metrics to In-Situ Discover Planktonic Interspecies Variations or Mutations

Abstract—Planktons are vital to our planet’s ecosystems. Mapping and monitoring planktonic-distribution in the ocean is of great importance in understanding these ecosystems as well as gaining insights on the general health of our planet. The task of identifying different planktonic species and their concentrations is a critical yet challenging part of this endeavor. In this paper, we explore the utilization of one-shot classifiers on in-situ captured planktonic images. The similarity-based classification method used by the one-shot classification algorithms gives rise to various research opportunities outside of the typical classification paradigm. Exploring the development of a proficient plankton image classification framework that can determine how different or how similar two plankton specimens are, can inform us if they belong to the same species, if they are different variations of the same species or if we have encountered a completely unseen classes of plankton. This similarity is valuable information that assists in further mapping the diversity of the planktonic species. We further extend the Siamese one-shot classifier with few-shot classifier to improve the model performance. Empirical evaluations of the new extension yield promising results.

Anton Tolstonogov, Far Eastern Federal University, Russia

The Modular Approach for Underwater Vehicle Design

Abstract—The authors propose a new modular design of an underwater vehicle in the article. The core idea is the complete separation of an underwater vehicle on different self-sustaining modules with its control and power system. It is proposed that wireless technology will provide data and power transfer between modules. The approach might be useful for design low-cost underwater vehicles for surveying in shallow water (up to 100 meter depth) with different replaceable payload modules. The paper presents the principal design of modules and their joint.

The communication system’s design based on Bluetooth Mesh is presented. The experiment shows reliable data transmission upto 35mm in seawater and up to 100mm in seawater with plastic supports between module housings.

The power management system is presented and discussed in the paper. The paper presents its overall design and the wireless power transfer system as the most critical node. The last one aims to unite batteries of all vehicle modules in the one “virtual” battery. It is shown that power transfer efficiency is up to 84% on distance 10mm between planar coils due to implementing the magnetic resonance method.