The use of Remotely Operated Vehicles (ROVs) in exploring and building infrastructure in the ocean is expanding. ROVs are performing tasks underwater that would be difficult or impossible to do with human divers. These vehicles are being used in increasingly complicated and demanding environments that require improvements in the methods for controlling these vehicles. Currently, research into semi-autonomous control is being conducted to aide ROV pilots in compensating for environmental disturbances and unknown dynamics. To effectively implement semi-autonomous control, precise thrust forces must be elicited from the thrusters. This work discusses a low-level thruster controller that can be used as part of a semi- autonomous guidance, navigation and control system for a ROV. A thruster dynamics model describing the thrust force of a propeller-type underwater thruster was derived and implemented for the thruster on the Saab Seaeye Falcon ROV. The thruster dynamics model described is a quadratic equation that uses the propeller velocity to determine thrust force. This model includes a mechanism for compensation against the external motion of the thruster, such as occurs when the ROV moves through the water. Several experiments were performed to fully characterize the quadratic thruster dynamics model and test its ability to accurately predict thrust force based on a known ambient water velocity and propeller angular velocity. The drag force was calculated and removed from the force measurements to get the thrust force used in the model. The model coefficients were determined and then the resulting model was tested against experimental data to determine the efficacy of the model in the lab environment and compare it to a widely used linear thruster dynamics model. The results showed the quadratic model improved upon the linear model, and the quadratic model was valid over a larger range of ambient water velocities. The quadratic model was then inverted to provide a thruster control algorithm that determines the propeller angular velocity necessary to produce a desired thrust force. This algorithm was used to design a low-level thruster controller. This controller was designed to be used on an existing vehicle where thrust force feedback is not available and difficult or expensive to add. This allows it to be used in a wider range of applications than controllers that rely on such feedback to operate. The controller was implemented using a PID control loop to drive the angular velocity of the propeller to the desired rate. An iso-parametric mapping, which transforms the linear PID output to the non-linear thruster input, was added to provide a faster response time for the controller over the entire range of the propeller velocity. The performance of this low-level thruster controller was demonstrated in the test environment. The low-level thruster controller followed a desired thrust force under a range of ambient water velocities. The thruster characterization and low-level thruster controller was designed to be used on an existing ROV. The motivation behind this work is to build a controller that may be implemented for use by a high-level vehicle controller. The low-level thruster controller presented here does not depend on sensors or equipment that is largely unavailable on vehicles without costly retrofits, and the experimental characterization does not require intimate knowledge of the inner workings of the thruster. This makes it easy to implement and generalize to a variety of thrusters. The results of this work show a low-level thruster controller than can be used in a control schema for existing ROVs. / Graduate / 0547 / matt@amosbuchanan.net

This project had two primary goals: (1) to explore opportunities to further a deep-ocean vehicle’s reach using alternative pressure spheres, and (2) to implement an existing deep-ocean vehicle (lander) in active scientific research. I gained a greater understanding of the limitations and design choices made for existing pressure spheres using Finite Element Analysis (FEA). My simplified FEA model predicted sphere failure for the existing 30% Fiber Glass 70% Nylon injection molded spheres at an external pressure of 3,954psi or 2,690m ocean-depth (only a 7.38% error compared to the tested minimum failure depth), so I determined it a valid model. I also explored alternative designs and materials that could be used for pressure spheres in deep-sea applications. Existing pressure sphere models filled with an incompressible fluid failed at 12,670psi or 8,621m ocean-depth - over three times the depth of the same sphere filled with air. Next, I varied the sphere thickness of existing spheres to determine its impact on depth rating. While the increased thickness did provide an increase in depth rating, there were diminishing returns as the sphere was made thicker. I deemed both of these design options infeasible for our application. To consider the use of laminated composite spheres, the addition of an equatorial ring was required to manufacture O-ring seals safely and reliably. A simple cylindrical equatorial ring model using a stainless-steel ring had a predicted failure at 3,017psi or 2,053m ocean-depth. While this model predicted failure at 637m shallower than the sphere without the ring, it was the only ring material tested to reach the rated depth for the existing pressure spheres (2km), so I concluded stainless-steel is the best ring material. A spherical stainless-steel equatorial ring design was then analyzed which predicted failure at 3,915psi or 2,664m ocean-depth – only 8.3% less than the original sphere with no ring. Because of its successful performance and near identical results to the original model, I determined a stainless-steel spherical equatorial ring is the best option for laminated composite sphere sealing. Finally, I analyzed three different kinds of laminated composite pressure spheres: two carbon fiber and one fiber glass. Each laminate was designed to be quasi-isotropic and as close to 0.8” thick as possible to keep it consistent with the original sphere design. The sphere made of 584 Carbon Fiber with a lay-up of: [[-45/45/0/90]6]s was found to predict failure at 10,000psi or 6,804m ocean-depth, more than 2.5 times that of the original sphere. Next, a model made of 282 Carbon Fiber with a lay-up of: [[-45/45/0/90]11]s predicted failure at 9,242psi or 6,289m ocean-depth – more than 2.3 times as deep as the original pressure spheres. Lastly, a sphere of 7781 Fiber Glass with a lay-up of: [[-45/45/0/90]11]s predicted failure at 6,630psi or 4,511m ocean-depth – about two-thirds the depth of the 584 Carbon Fiber composite, but more than 1.6 times the depth of the original sphere. While real-life applications of these materials would include design modifications and manufacturing imperfections which would lower their maximum depth rating, these results are highly encouraging and show that all three materials could be viable options for future production. Additionally, through partnership with Dr. Crow White and his marine science undergraduate students, I completed numerous deployments for a Before and After Controlled Impact (BACI) study on the area of the proposed windfarm off the coast of Morro Bay, CA. Many modifications were made to the existing lander which enabled it to successfully be implemented in these studies including a new bait containment unit, light color filters, a GPS tracking device, and a large vessel recovery device. A total of 5 pier deployments and 3 boat deployments were conducted by my team over the course of 6-months. Planning for these deployments included accounting for budgeting, weather, permitting, and multi-organizational logistics while working with both NOAA and the Cal Poly marine operations staff.