The Student Poster Competition at OCEANS 2019 Seattle

Shyam Madhusudhana, OES Student Poster Competition Chair,
Photos by Stan Chamberlain

The Student Poster Competition (SPC) is a flagship event of the MTS/OES OCEANS conferences in which students (both undergraduate and graduate) from colleges and universities around the world participate. Col. Norman Miller had created and implemented the first SPC in 1989 during the OCEANS conference in Seattle, Washington, and it has been a feature of OCEANS conferences ever since. It was befitting that the 30th anniversary of SPC had returned to Seattle. During each conference, typically about twenty students are selected from the pool of aspiring applicants, based on their submitted abstracts, to present a poster describing their work. The selected students’ travel, accommodation and conference registration are fully supported by the conference. In addition, monetary prizes are also awarded to the top three best presenters among the selected participants.

The second of this year’s SPCs was held during OCEANS 2019 Seattle at the Washington State Convention Center from the 27th to the 31th of October. The SPC program was funded by a grant from the Office of Naval Research (ONR), and we offer our sincere gratitude for their continued support. Preparations in the run-up to the event were ably managed by the local organizing committee (LOC), and in particular by Miles Logsdon who was the SPC chair on the LOC. Out of the 20 applicants that were shortlisted, four candidates had to withdraw due to different reasons, and we had 16 students presenting their work at the event.

The students put up their posters at designated slots on the Skybridge on October 28th. In this edition of the SPC, we had a total of 18 volunteers that offered to serve as judges. We understand that this commitment was an imposition on the judges’ conference schedules, and we deeply appreciate their dedication to encouraging the next generation of ocean scientists and engineers. To ease the load on students as well as the judges, the judging process was restricted to four short sessions spread over two days.

The participants and the judges were all gathered at the main stage on the morning of November 30th. Each participating student was awarded a Certificate of Attendance, and the judges’ efforts were acknowledged by presenting each of them a Certificate of Appreciation. The three top-ranked posters were announced and prizes were awarded.

The full list of 16 participants (including the prize winners) together with their affiliation, poster title and an abstract of their poster are given below.

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

Bernardo Teixeira, INESC TEC, Portugal

Deep Learning Approaches Assessment for Underwater Scene Understanding and Egomotion Estimation

Abstract—This paper address the use of deep learning approaches for visual based navigation in confined underwater environments. State-of-the-art algorithms have shown the tremendous potential deep learning architectures can have for visual navigation implementations, though they are still mostly outperformed by classical feature-based techniques. In this work, we apply current state-of-the-art deep learning methods for visual-based robot navigation to the more challenging underwater environment, providing both an underwater visual dataset acquired in real operational mission scenarios and an assessment of state-of-the-art algorithms on the underwater context. We extend current work by proposing a novel pose optimization architecture for the purpose of correcting visual odometry estimate drift using a Visual-Inertial fusion network, consisted of a neural network architecture anchored on an Inertial supervision learning scheme. Our Visual-Inertial Fusion Network was shown to improve results an average of 50% for trajectory estimates, also producing more visually consistent trajectory estimates for both our underwater application scenarios.

Second Prize (Certificate and $2000 prize)

Philip Leadbitter, University of East Anglia, United Kingdom

A methodology for Thorpe scaling 512 Hz fast thermistor data from buoyancy-driven gliders to estimate turbulent kinetic energy dissipation rates in the ocean

Abstract—A Kongsberg Seaglider with a microstructure package was deployed in the Faroe-Shetland Channel in 2017 as part of the 4th Marine Autonomous Systems in Support of Marine Observations (MASSMO4). Using the FP07 fast thermistor (512Hz), the standard Seaglider thermistor (0.2 Hz) and potential density calculated from Seaglider conductivity-temperature sail (0.2 Hz) a comparison of the Thorpe Scale method has been made. Through this method turbulent kinetic energy (TKE) dissipation rates are inferred from the length-scale of a turbulent overturn. Comparison of the three physical quantities show that overturns with a comparable length-scale also have a comparable TKE dissipation rate. The range of estimated TKE dissipation rates from the 0.2 Hz data is also comparable to those inferred using the same method applied to potential density calculated from a ship mounted CTD.

Third Prize (Certificate and $1000 prize)

Nare Karapetyan, University of South Carolina, USA

Dynamic Autonomous Surface Vehicle Control and Applications in Environmental Monitoring

Abstract—This paper addresses the problem of robotic operations in the presence of adversarial forces. We presents a complete framework for survey operations: waypoint generation, modelling of forces and tuning the control. In many applications of environmental monitoring, search and exploration, and bathymetric mapping, the vehicle has to traverse in straight lines parallel to each other, ensuring there are no gaps and no redundant coverage. During operations with an Autonomous Surface Vehicle (ASV) however, the presence of wind and/or currents produces external forces acting on the vehicle which quite often divert it from its intended path. Similar issues have been encountered during aerial or underwater operations. By measuring these phenomena, wind and current, and modelling their impact on the vessel, actions can be taken to alleviate their effect and ensure the correct trajectory is followed.

Hugo Antunes, University of Porto, Portugal

Autonomous Identification and Tracking of Thermoclines with a Vertical Profiler using Extremum Seeking Control

Abstract—The thermocline is a relatively narrow vertical region that separates the mixed layer at the surface from the deep-water layer. In this region, the gradient of temperature with respect to depth is higher than in the rest of the water column. The characteristics of the thermocline have strong impact in marine biology, since it may trap high-nutrient organisms, and it also affects sound propagation, with direct impact on underwater acoustic communications and military operations. Under adaptive sampling, Autonomous Underwater Vehicles are practical tools for efficient ocean observation. In this work, we describe an implementation of an Extremum Seeking Controller that performs identification and tracking of thermoclines at its point of highest temperature gradient in a completely autonomous way. The vehicle chosen to perform this tracking was an autonomous vertical profiler, and the algorithms were validated using both real and simulated data.

Xinwei Chen, Memorial University of Newfoundland, Canada

Automatic Identification of Rain-contaminated Regions in X-band Marine Radar Images

Abstract—A self-organizing map (SOM) based method for identifying rain-contaminated regions in X-band marine radar images is proposed. The difference of texture and pixel intensity distribution between rain-contaminated and rain-free echoes is first exploited. A Gabor filter bank is designed to filter marine radar images and generate texture features. Bin values extracted from the localized histogram can represent pixel intensity features. Both types of features extracted from each pixel are combined into a feature vector and trained using an unsupervised neural network, SOM, which clusters pixels into rain-free and rain-contaminated types. Images collected from a shipborne marine radar in a sea trial off the east coast of Canada under rain conditions are utilized to validate the proposed method. Identification results generated from clustering show that the rain-contaminated pixels are effectively detected.

Vincent Dupont, Université Laval, Canada

A region growing algorithm adapted to bathymetric point clouds

Abstract—This project focuses on the extraction of plane surfaces and their uncertainty in a bathymetric point cloud. To do so, we propose the Region-Growing-Adapted-to-Bathymetry (RGAB) algorithm. Knowing where plane surfaces on the sea-bottom are is needed in calibration of MBES system. These surfaces are also involved in the computation of error estimators used to validate data acquired by ASV. With the RGAB algorithm, we propose several adaptations to classic region growing to be used with bathymetric point cloud. Indeed, we use a more robust PCA approach to extract local normal vectors and a raster-based analysis on the bathymetric DTM to get seed regions for all plane surfaces.

Wonse Jo, Purdue University, USA

Development of an Unmanned Surface Vehicle for Harmful Algae Removal

Abstract—In this paper, we introduce a small and low-cost unmanned surface vehicle (USV), the SMARTBoat 5, capable of removing harmful algal blooms (HABs), which are a rising environmental issue worldwide. The developed USV is a hovercraft type, operated by two propellers with duct fans; it is able to freely move even in shallow water and to approach shorelines. For eco-friendly, immediate, and safe control of algae, the USV is equipped with a novel water suction mechanism that enables it to actively collect algae without physical contact. In addition, it is equipped with a mesh net-based algae filter system that is easily disassembled and replaced. The USV system is supported by the Robot Operating System (ROS) for expandability and use in diverse applications. The performance of the proposed water suction mechanism and USV platform overall are validated through computational fluid simulation (CFD) and experiments in both lab and real environments.

Patrick Kelley, Lake Superior State University, USA

Detection of Oil Under Ice: An Acoustic Reverberation Method

Abstract—Through new exploration and aging infrastructure, there is an ever-increasing risk of oil spills or leaks into natural bodies of water, but presently there is no low-cost method to actively monitor for such spills under ice cover, particularly not any method giving rapid coverage of a large area. Therefore, a method based on acoustic reverberation time under the ice was investigated. A broadband, impulsive acoustic source generating a frequency band of 10–90 kHz was used in a water tank with a layer of naturally-grown ice for a scaled feasibility study of this method; this band provided appropriately short wavelengths, scaled for the small tank environment. Reflections from the tank walls were minimized, and a diffuse environment within the tank allowed for extraction of a reverberation time with a hydrophone. Results show an oil presence changed the reverberation time as compared to an oil-free ice layer across all diffuse frequency bands.

Kevin Mack, Clarkson University, USA

Time-of-Flight (ToF) Cameras for Underwater Situational Awareness

Abstract—Recently developed commercial Time-of-Flight(ToF) cameras have been used to accurately and reliably measure scene depth with high resolution in applications such as automotive LiDAR. There is a desire to adapt this technology for applications in underwater environments. In this work, we establish a methodology for using modified commercial ToF cameras in turbid water. We express the need for hardware and software modifications to the camera and demonstrate initial results in the efficacy of the camera in an underwater test scenario. We include ToF camera imagery taken under a variety of water conditions to understand the performance limitations of this technology as a function of water clarity. Target detection results from preliminary laboratory test tank experiments are presented for two different classifiers, each of which achieves high accuracy for a certain range of water conditions.

Curran Meek, University of Hawaii at Manoa, USA

Fast Autonomous Underwater Exploration using a Hybrid Focus Model with Semantic Representation

Abstract—By combining complex exteroceptive perceptions, humans are able to seamlessly explore a new environment, build up their knowledge base, and “easily” identify novel objects or imminent hazards. This work aims at employing a similar intelligent cognition and exploration system for unmanned marine robots. The proposed system semantically discriminates the visual inputs of the robot’s environment by combining task-directed and stimulation-based focus to encode artificial “attention”, to enhance exploration decision making. Additionally, the hybrid focus model can use the semantic descriptors to label a target object (e.g., a certain type of coral) with metric information (e.g., geographical locations of the coral) from a range-and-bearing sensor, allowing a metric map to be generated with semantic labels of the desired types of objects. These maps can help simplify human and robot interaction by providing a clear semantic interpretation of the robots observations. The proposed system is implemented and tested on an indoor robot and with underwater images.

Tamzidul Mina, Purdue University, USA

A Novel Double Layered Weighted Potential Field Framework for Multi-USV Navigation towards Dynamic Obstacle Avoidance in a Constrained Maritime Environment

Abstract—Motion planning of multiple unmanned surface vehicles (USVs) towards increased autonomy and wider coverage of the maritime environment is a pertinent requirement. Given the numerous types of USVs currently available with a wide spectrum of maneuvering capabilities, we present a generalized multi-USV navigation framework adaptable to specific USV maneuvering response capabilities for dynamic obstacle avoidance. The present paper integrates an optimal path planning with safety distance constrained A* algorithm and a proposed adaptively weighted potential field based path following approach with collision avoidance based on USV maneuvering response times. The system allows USVs with fast maneuvering abilities to react late and slow USVs to react sooner to oncoming moving obstacles gradually such that a smooth path is followed by the USV group with reduced cross track error. Simulation results validate reduced cross track error for slow and fast maneuvering response time multi-USV teams.

Benjamin Murphy, Dalhousie University, Canada

Lab-on-a-chip sensor for in situ nutrient monitoring

Abstract—Presented here is a low-cost in situ nutrient sensor based on lab-on-a-chip (LOC) technologies. A microfluidic chip with integrated check valves and a novel absorbance cell forms the backbone of the sensor. Using spectrophotometry to measure concentrations of nitrite and phosphate in liquid samples, effective limits of detection (LOD) of 30 nM and 100 nM are achieved for nitrite and phosphate respectively using a 10 mm optical path. The sensor was deployed in the Bedford Basin in Halifax, Canada for 12 hours for in situ phosphate measurement and analysis.

António Oliva Afonso, University of Porto, Portugal

Underwater Object Recognition: A Domain-Adaption Methodology of Machine Learning Classifiers

Abstract—This paper presents a novel dataset, composed of images of objects in two distinct environments and both controlled and uncontrolled capture conditions, aimed at serving as a benchmark for domain-adaptation image classification algorithms in an air versus underwater context. All images are fully annotated, extending the use of the dataset for detection as well as segmentation tasks. An exemplifying use-case is tested, where the performance of a Support Vector Machine applied to a Bag-of-Visual-Words and SIFT features is evaluated on both domains, with different training methodologies. Results demonstrate that the conventional classifier used has a lack of generalization ability, with a poor transfer of knowledge from the aerial to the aquatic domain.

Pedro Oliveira, University of Porto, Portugal

Experimental evaluation of segmentation algorithms for corner detection in sonar images

Abstract—Corners usually appear very distinct from the rest of the scene in a mechanical scanning imaging sonar (MSIS) image, generally characterized by sharp intensities. The detection of corners is particularly useful in human-structured environments such as tanks because the knowledge on their location provides a way to compute the vehicle position. The combination of some basic operations typically used for image segmentation have great potential to detect and localize corners in sonar images automatically. This article proposes and evaluates with experimental data a set of image segmentation algorithms for corner detection in sonar scans. The developed algorithms are evaluated with ground truth, and their performance is analyzed following a few relevant metrics for autonomous navigation.

Yu Herng Tan, National University of Singapore, Singapore

Design of a Morphable Multirotor Aerial-Aquatic Vehicle

Abstract—The design of any aerial-aquatic vehicle faces the unique challenge of vastly different fluid dynamics of air and water. Due to the demands of aerial propulsion, most existing solutions are fundamentally aerial vehicles with an auxiliary capability to operate underwater. Instead of using standard aerial platforms directly, we propose a novel morphable design for a multirotor-based vehicle inspired by the thruster layouts of regular aerial quadrotors and underwater remotely operated vehicles (ROVs). By adding the ability to rotate each motor about its respective arm, the proposed mechanical linkage allows for direct actuation in all six translation and rotational directions. The proposed concept is demonstrated in a proof of concept prototype and the details of the design and physical implementation are shown in this paper.

Alejandro Torres, Florida International University, USA

Correlating Water Quality and Profile Data in the Florida Keys

Abstract—Aquatic ecosystems present complex structures susceptible to changes that can cause adverse effects in the water. These problems have turned the attention of researchers to understand them, and possibly take action to prevent further damages. This interest led to the accumulation of large amounts of data with limited personnel and resources to analyze it. An example of this is the collection of data in South Florida for 25 years by Florida International University. By making use of a depth profile and surface water quality data sets collected in the same location at the same time, a methodology is proposed to correlate these two data sets. By using Machine Learning, we represented depth profiles with coefficients followed by clustering analysis. Similarly, a water surface chemical dataset was clustered using k-means. We then used statistical methods to test the connection between these two data sets.