My research in PhD invloved the exploration of a new method for sampling environmental variables in oceans using autonomous robots. It is underwater collective sensing. The popular methods used for positioning underwarer are based on dropping buoys into the ocean and recording the location of them using satellites. The buoys are displaced by the currents at a particular depth and when they emerge to surface, their location is recorded and the cycle is repeated for a few hundred times. The data is then graphed to show the underwater current as interconnected straight lines.
The downfall with this method is that movement underwater is ignored and is assumed that the buoys move in a straight line from each recorded location to the next. However this is unlikely.
For applications that need a higher resolution, there is a demand for a better approach. The new method, called Flock Distortion, is suitable to find underwater currents in much more detail. This is based on detecting disturbances caused on a flock of autonomous robots, which can be detected by the changes applied to the distances between each pair of robots. This disturbance is interpreted to show what the original current profiles were, along with the sensory information collected during the sampling of the ocean. The data can then be plotted and used by meteorologists, life scientists, or for any other applications.
For my Ph.D. I worked with my Prof. Chris Melhuish and Prof. David McFarland. Ian Gilhespy helped us on the engineering aspects of the project.
Ph.D. Abstract
Currently no equivalent system of GPS exists for underwater applications. There is therefore a need to track objects underwater by other means. In this thesis a novel algorithm is presented which employs a flock of underwater robots to generate depth profiles without the aid of marker communication systems. In this approach only interrobot distances are required.
Here, the sampling flock can map over long distances and is not constrained to the marker communication range. This thesis demonstrates a method for calculating the positions of a flock of underwater robots using only the recorded relative positional information of the robot flock. The calculated absolute path of the underwater robots can then be used to map the sampled sensory information such as CTD (Conductivity, Temperature, and Depth) in 3D. The flock is dropped into the ocean and is not tracked by any external device until it resurfaces later on. The data is collected from the ‘surviving’ robots and a novel algorithm is applied to this data to calculate the absolute 3D trajectory taken by the robots over the duration of the sampling. In this thesis, the flock estimation algorithm and the associated issues of generating a velocity profile are presented and discussed.
In conjunction with this research a prototype underwater robot is also developed which is capable of altering its density by heating oil. It is designed to gather information in the ocean, as a member of a flock of such robots. The Divebot’s mechanical and electronics design is explained and a detailed mathematical model of the physics of Divebot is developed. This is then used to simulate a flock of Divebots that are then used with flock estimation algorithm to verify the algorithm’s ability in estimating the trajectories of the robots. You can download the thesis below.