Abstract:
A Remotely Operated Vehicle (ROV) assists in performing missions underwater.
Equipment such as cameras, lights, and robotic arms are assembled on the
ROV to help complete the required task. In this thesis, the motivation behind the
mission of the ROV is to monitor sea turtles underwater by embedding a camera
on the ROV to track their behavior. A complete coverage path is proposed for
the ROV to follow, and various controllers are designed to ensure that the ROV
tracks the reference path.
In the first part, this thesis describes a complete Coverage path planning
(CPP) methodology for an ROV. The proposed algorithm relies on a priori knowledge
of the 2D map environment, which may contain obstacles. The approach
combines Boustrophedon Cellular Decomposition (BCD) with a convex decomposition
method to allow handling complex areas. After decomposing the area of
interest into several convex subareas, different path coverage designs are investigated
to find the optimal coverage path. The path consists of many patterns such
as straight line segments, right-handed arcs, left-handed arcs, and zigzag curves.
Energy, time, coverage rate, path length, and the ability of the controlled ROV
to follow the desired path as measured via position error are the criteria considered
to design the optimal path. Results show that using the right-handed and
left-handed curves in the path design decreases the energy consumed and thus
results in the optimal coverage path. In the second part, this thesis considers
the control problem. The control scheme is divided into two main parts: control
allocation and controller design. The control allocation problem is based on
having more actuators than needed as the ROV has four horizontal thrusts that
allow the motion in only three Degrees of Freedom (DOF). For the control allocation
problem, analytical and numerical methods are combined to compute the
actuator input required to achieve the given desired forces and moments. For the
controller synthesis problem, a classical Proportional-Integral-Derivative (PID) controller is first designed and used to benchmark the performance of other more
sophisticated controllers. This PID controller is the one used for determining the
optimal path in the first part of the thesis. A back-stepping controller and an
H-inf robust controller are then designed and compared. The simulation results
show that the H-inf robust controller provides better path tracking results than the
PID and back-stepping controllers when applied to the nonlinear model of the
ROV. The robustness of the designed controllers is inspected by incorporating
the effect of underwater currents in the simulations. Future work would focus on
integrating the camera on the ROV and conducting real-life experiments in the
Mediterranean Sea next to the shores of Lebanon to track the sea turtles.