Getting Started

Here is an overview to help you build configure and run the Towed ROV

MECHANICAL SECTION

This section explains the mechanical components of the ROV.

Hydrodynamics

Wings are the primary control surfaces of the Towed-ROV generating the necessary lift to maintain or achieve the desired depth. This guide can be used an aid in designing the lifting surface of the ROV.
Hydrodynamic lift: As a lifting element or foil (wings in this case) pierces a fluid at an angle relative to the flow, it creates a differential pressure between the top and bottom surfaces of the element. This pressure difference when integrated over the entire surface sums up to a net force called "Lift". The direction of the force depends on the orientation of the lifting element.
Angle of attack: The relative angle between the lifting element's center line and direction of flow

Span: The dimension of the lifting surface in the direction perpendicular to flow.
Chord: The dimension of the lifting surface in the direction of flow
Stall angle: The lift force of a foil increases with the angle of attack until a critical angle - called stall angle. Beyond stall angle, the lift generation capacity of a foil starts to decrease.

Aspect ratio: The ratio of span to chord is called 'aspect ratio'. It is a very important design parameter in fluid dynamics. A lower aspect ratio (for a given wing area) improves the stalling range and therefore results in a higher achievable max. lift. But, at the same time it also increases the lift-induced drag. Similarly, a higher aspect ratio foil, though theoretically results in a higher lift coefficient, it stall prematurely and therefore has a lower achievable max. lift force. The stall performance and hence the lift generating capacity of a lower aspect ration foils seems to be better than its high AR counterpart. However, there will be a significant lift-induced drag in a low AR foil. For this reason, a high AR foil is preferred.
Design rule 1: Maintain high AR i.e., foil span > foil chord
Torque due to lift: The hydrodynamic lift force acts at an average distance of approx. 0.3c to 0.4c from the leading edge of the foil, where 'c' is the foil chord. Considering the average value of 0.35c for the foil axis, the maximum arm is 0.05c.
The torque due to lift is therefore, F * 0.05c .
Viscous drag: It is an another component of the hydrodynamic forces acting on the foil due to its movement in the fluid. While the lift force acts normal to the surface, drag force acts parallel to it. At low angles of attack, viscous or skin drag is the dominant component and can be reduced by selecting more streamlined profiles (unlike a flat plate) Profile geometries: Much of the discussion stated above is based on flat plate wing profiles. Though flat plates have a relatively low lift performance compared to many other streamlines geometries (like NACA profiles), their simple geometry and mathematical treatment makes them easy choices in prototype stages.

Online calculator:

A simple online calculator is made to aid the used in quick design of Towed ROV wings. The mathematical relations used in the calculator are based on literature available in similar applications in aerospace and marine domains. Therefore, they only provide a reasonable rough estimate of forces.
The online calculator can be found here

CFD Simulations

From the forces and moments calculations we can see that the correct estimation of drag force and its point of application are critical in solving the force equation and finding the optimum location for placing the towing point.
Net drag force on the ROV is a component of the hydrodynamic forces. It is usually derived by solving a set of non-linear partial differential equations called Navier Stokes which are solved by CFD solver packages.
For this project, we used SimScale. It is a browser based freemium tool, which can be easily opened from a browser and used for free (with limited computational resources that are sufficient for small scale usage).
We explain the geometric model, boundary conditions, meshing and simulation settings as below.
Below is the simplified model used for the flow simulation.

Note: The model used for simulation is different from the final assembly which has more frame members for mounting other components.

Bounding box: It is the volume of fluid which is of interest in the simulation because it is not realistic to consider the whole fluid volume (in our case, the water body like lake or sea).

Boundary conditions: Next important step is to input the boundary conditions. For this simulation, the below are used.
Inlet velocity - 2m/s
Outlet pressure - 1 bar
ROV body - Wall no-slip
Other faces - Wall slip
Meshing: A bi-resolution mesh boundary is defined with a coarse mesh on the outside with fine mesh closer to the ROV. This is to save on the calculation time without compromising too-much on the accuracy.

Results:
Velocity field

The force and moment plots are shown as below
Horizontal drag - 160 N
Vertical drag - 30 N
Transverse drag - 0 N (approx)

Structural Design

With inspiration from the exposed skeleton structure of quadcopters came the idea of using standardized aluminium profiles to create a open modular frame. This gave the project potential for rapid modification of frame structure and placement of the necessary components for pitch control

Modular Structure Profiles

The groved aluminum profiles make the structure modular and can be adjusted according to any dimension.

Locking Nuts

The locking nuts enable to attach payloads and mountings at any suitable place on the ROV.

Pitch Mechanism

A pitch mechanism is developed that could incorporate the frame mounting brackets in the design. Due to uncertainty of material properties of having oil on one side and seawater on the other side of a PLA plastic wall of 8mm, some solutions were proposed.

• The pitch mechanism has to be printed with 100% to remove the air inside the walls
• PLA could be exchanged for PETG due to its seemingly stronger waterresistance.
• A form of coating could be applied to the pitch housing. This could be a form of marine coating or epoxy.

By removing the utility plate and moving the necessary components closer to the towing point on top of the ROV, the frame design was adjusted to compress the overall height. The aluminium profiles were held together by the side walls that were made from acrylic. These walls cracked during the assembly and were considered a problem. Due to their fragility, it was deemed necessary to remove them. This led to designing corner brackets out of PLA. Simmilar brackets had already been tested during sea trail as four brackets were used to hold the utility plate in place.

Mountings

We designed 3D printed mountings for the ROV

Bracket 1

Bracket 2

Sonar Mounting

Download Files

ELECTRONICS

We have enclosed all the electronic components in a water proof container by Blue Robotics. This way we can play more with the design of the ROV making sure that our electronics are safe.

Hardware Components

Here is the list of components/electronics used in this project

Schematics

You can download the Schematics of the electronics circuits

Download Files

SOFTWARE

This section explains the software component of the ROV

Graphical User Interface

An interactive GUI is made to control, send commands and send/receive data to and from the ROV. Our GUI is developed in a javascript frontend framework React (Chakra UI)

The GUI can be installed and run using the following commands

git clone https://github.com/Towed-ROV/gui.git   

cd gui  

npm install  

npm run start   

API

We have used FAST API for the data transfer between between the GUI, database and payloads on the ROV.

The API can be installed and run using the following commands

git clone https://github.com/Towed-ROV/api.git   

cd api/control-api  

python -m venv venv

venv\Scripts\activate

pip install -r requirements.txt  

Then run the API using the following commands

cd app  

uvicorn main:app --reload   

Trash Recognition

One of the tasks of the ROV is to stream / record videos of the ocean floor to find possible plastic waste or fishing gear. To tackle this challenge we used a machine learning model to detect and classify the plastic waste. We are not only concerned with “there’s something” but also need to know “what are we seeing”. This problem is a classification problem and since we can have multiple objects of the same or different type of the objects, we need to know which pixel corresponds to each object

The data set was highly relevant to our use case and is published by a Japanese Research Organization JAMSTEC. The dataset contains 23 classes of objects underwater.

The model used is Mask RCNN model by matterport. This model is trained on the COCO dataset by Microsoft. We used this model as our base and further built our model on top of it and performed annotation conversions as well as model configurations.

PAYLOADS

We can attach various payloads to the ROV according to our requirements.

Side Scan Sonar

One of our payloads is the side scan sonar. The side scan sonar runs with an API from DeepVision. The side scan sonar gets a broad view scan of the ocean floor. The API provided by Deepvision does not provide real-time results, we have integrated it in our code to show real time result in our GUI.

**This page is copied from the webpage on github