Dear Members of the Selection Committee,
I am writing in connection with the opportunity of taking part in the multidisciplinary PhD program at IMT in Lucca as it has a well-known and excellent reputation . I would like to apply for this position because I believe that my academic education and experience are well suited for the needs of this endeavor.
I am currently working at a company named A.T.S. (Advanced Technology Solutions) in Catania. Even though my main duties are not related to research and development, recently I was involved in several research projects promoted and funded by the European Union and the region of Sicily. The most challenging one is regarding Field Force Management (FFM) in which I am currently working with distinguished partners such as Enel, the University of Salerno, and Opera 21. Basically, the aim of the project is to improve and optimize the maintenance operations in electrical energy distribution plants and in renewable energy plants (especially photovoltaic plants). My responsibilities range from the definition of novel ontologies, which are exploited by innovative web-applications, to the implementation of multi-objective optimization algorithms. Particularly regarding the latter, the results obtained from Tabu Search and Simulated Annealing implementations are taken into account and compared in order to solve the problem of human resources scheduling and assignment for different daily tasks.
In contrast, my academic research activity was related to the study of control solutions for Unmanned Aerial Vehicles (UAV) and this is the research study which I would like to pursue at IMT. As my references show, professor XXXX was my supervisor for both my bachelor’s degree and master’s degree projects.
Specifically, my bachelor's project was an integrant part of a project called VOLCAN,which has been pursued by professor XXXX in this decade at University of Catania. The aim of the entire project was to analyze the gas composition of volcanic clouds. Obviously, such an operation is dangerous and almost inefficient if performed by human beings. Indeed, this kind of study should be executed before that volcanic gases combine themselves with the atmosphere because the experimental data would consequently be altered. Having said this, an UAV could be considered almost an essential tool in order to accomplish such an operation. My job consisted of actualizing software architecture in Matlab-Simulink that is able to interact with the popular flight simulator X-plane. The main task was to obtain the automatic tuning of a PID controller parameters for the roll movement of the aircraft starting from dynamics of ailerons. At the end, the project unveiled an useful tool for tests in the field. In fact, the principal advantage gained from such software architecture was that it allowed us to save a large amount of time which could otherwise have been spent in the manual tuning of the controllers during flight tests.
Moreover, I concluded my academic education with my master project which was also related to VOLCAN, but its scope was significantly different. The cooperation between UAVs and UGVs (Unmanned Ground Vehicles) was the principal goal to achieve because it would have lead to better results in terms of path building and target detection. Basically, this kind of cooperation shows its strength and efficiency during scientific missions on the volcano in order to collect data as well as during rescue or civil surveillance missions. My responsibilities consisted of validating experimentally a Kalman filter based algorithm which allowed us to estimate with high precision the position of a fixed target on the ground starting from the noisy measurements provided by GPS and AHRS sensors mounted on the aircraft. This validation was executed by implementing the novel algorithm in a Delphi application capable of interacting with Google Earth. This software environment was created in order to exploit the real data acquired from instruments and collocate a virtual monocular camera on the world gaining, consequently, a significant advantage before performing tests in the field. Fundamentally, the vision algorithm was essential in order to build an approximate model of the three-dimensional world starting from two-dimensional images, while the extended Kalman filter was designed expressly to reduce the consequent uncertainty of the results derived from the visual reconstruction process and from the noisy measurements.
The purpose of my research plan that I would like to pursue at IMT is strictly related to my academic research experience and consists of improving the geo-localization performances for a UAV flying in a no-structured environment. Moreover, I believe that, in some applications, a policy of searching targets would be crucial in order to minimize the flight path in terms of distance and time. For these reasons, it would be useful to build a simulation platform for the implementation and the consequent validation of different well-known algorithms in order to solve the searching targets issue. In addition to this, the majority of real applications need the identification of a moving target on the ground which might be a person or a car, for example, since targets do not often move in a known and structured environment.
Therefore, a novel approach based on semantic SLAM (Simultaneous Localization And Mapping) would be helpful if applied to UAV control systems. Particularly, the term “semantic” is appropriate because the algorithm would be able to associate fixed features on the ground (exploited like landmarks) with known objects by giving them meaning. Likely, this approach would significantly reduce the computational burden in those cases where the camera provides information on fixed features that were relevant in other past images during the flight. Fundamentally, the control system would be able to “recognize” a feature already captured in the past and exploit that information to geo-localize targets quickly and precisely. In real applications such as traffic monitoring or civil surveillance, this kind of occurrences is very frequent because UAVs often have to go through a predefined path, therefore this new approach would probably be applied widely.
As you may have noticed, I applied for the “Control Systems” and for the “Image Analysis” curricula because, despite their objective differences, there is a strong relationship between them. Indeed, computer vision and control algorithms are often well integrated in many SLAM applications.
It would be an honor to discuss with you my resume either personally or by a Skype interview. I am looking forward to hopefully hearing from you.
Thank you for your time and your consideration.