| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.89 MB | Adobe PDF |
Authors
Abstract(s)
O amplo espaço marítimo Português p ermite o desenvolvimento de diferentes atividades, tais como o comércio internacional, a p esca e o turismo. Porém, a
o corrência de incidentes marítimos p o de aumentar e ameaçar a p opulação. A p oluição das águas, o aparecimento de caravelas-p ortuguesas nas praias, os acidentes
com embarcaçõ es e a queda de arribas, constituem algu ns desses incidentes. Como
tal, torna-se imp erativo a vigilância e monitorização destas ameaças p or parte das
autoridades marítimas.
Assim, de forma a auxiliar as autoridades marítimas na vigilância e monitorização destas ameaças, a presente dissertação prop õ e u m sistema de ap oio à decisão
de baixo custo comp osto p or 4 comp onentes: veículo aéreo não-tripulado, sensor,
aprendizagem automática e interface gráfica. Este sistema tem como ob jetivo detetar, identificar e classificar in cidentes marítimos nas zonas costeiras p ortuguesas em
temp o real contribuindo para uma melhor eficácia das autoridades marítimas. Para
isto, foi testado o uso de um veículo aéreo não-tripulado, de p equenas dimensõ es e
baixo custo, com sensor ótico na recolh a de imagens de manch as de hidro carb onetos
na água em ambiente simulado. Foram aplicadas fu ncionalidades de aprendizagem
automática do software Orange Data Mining To ol para treino e teste do reconhecimento e classifi cação das imagens com manch as de hidro carb onetos na água recolhidas. Posteriormente, foi criado um protótip o de uma interface gráfico de uma
aplicação móvel com recurso ao software Balsamiq. Este oferece a p ossi bilidade de
planear e controlar o vo o do drone e a receção de alertas de in ci dentes marítimos e
foi validado através do critério System Usability Scale.
Os resultados obtidos em ambiente simul ad o su gerem que o drone, o sensor
ótico e a aprendizagem automática p o derão constituir a solução d e um sistema
de ap oio à deteção de incidentes marítimos. Desta forma, este estudo contribui
para o auxílio das autoridades marítimas na deteção, identificação e classificação de
incidentes marítimos nas zonas costeiras p ortuguesas.
The ample Portuguese maritime space allows the development of different activities, such as international trade, fishing and touri sm. However, the o ccurrence of maritime in cidents can increase and threaten the p opulation. Water p ollution, the app earance of Portuguese caravels on b eaches, accidents with b oats and falling cliffs, are s ome of these incidents. As su ch, it b ecomes imp erative for maritime authorities to b e vigilant and monitor these threats. Therefore, in order to assist the maritime authorities in the surveillance and monitoring of these threats, this dissertation prop oses a low cost decision supp ort system comp osed of 4 comp onents: unmanned aerial veh icle, sensor, automatic learning an d graphic interface. This system aims to detect, identify and classify maritime incidents in Portuguese coastal areas in real time, contributing to a b etter effectiveness of maritime auth ori ties. For this, it was tested the use of an unmanned aerial vehicle, of small dimensions and low cost, with optical sensor in the collection of images of hydro carb on stains in the water in a simulated environment. The Orange Data Mining To ol software’s automatic learning features were applied for training and testing the recognition and classification of the images of hydro carb on stains in the water collected. Subsequently, a prototyp e of a graphic interface for a mobile application was created using the Balsamiq software. This offers the p ossibility to plan an d control the drone’s flight and th e reception of maritime incident alerts and was validated through the System Usability Scale criteria. The results obtained in a simulated environment suggest that the drone, the optical sensor and the automatic learning could constitute the solution of a system to supp ort the d etection of maritime i ncidents. Thus, this study contributes to the assistance of maritime authorities i n the detection, identification and classification of maritime incidents in Portuguese coastal areas.
The ample Portuguese maritime space allows the development of different activities, such as international trade, fishing and touri sm. However, the o ccurrence of maritime in cidents can increase and threaten the p opulation. Water p ollution, the app earance of Portuguese caravels on b eaches, accidents with b oats and falling cliffs, are s ome of these incidents. As su ch, it b ecomes imp erative for maritime authorities to b e vigilant and monitor these threats. Therefore, in order to assist the maritime authorities in the surveillance and monitoring of these threats, this dissertation prop oses a low cost decision supp ort system comp osed of 4 comp onents: unmanned aerial veh icle, sensor, automatic learning an d graphic interface. This system aims to detect, identify and classify maritime incidents in Portuguese coastal areas in real time, contributing to a b etter effectiveness of maritime auth ori ties. For this, it was tested the use of an unmanned aerial vehicle, of small dimensions and low cost, with optical sensor in the collection of images of hydro carb on stains in the water in a simulated environment. The Orange Data Mining To ol software’s automatic learning features were applied for training and testing the recognition and classification of the images of hydro carb on stains in the water collected. Subsequently, a prototyp e of a graphic interface for a mobile application was created using the Balsamiq software. This offers the p ossibility to plan an d control the drone’s flight and th e reception of maritime incident alerts and was validated through the System Usability Scale criteria. The results obtained in a simulated environment suggest that the drone, the optical sensor and the automatic learning could constitute the solution of a system to supp ort the d etection of maritime i ncidents. Thus, this study contributes to the assistance of maritime authorities i n the detection, identification and classification of maritime incidents in Portuguese coastal areas.
Description
Keywords
Deteção Remota Incidentes Marítimos Veículos Aéreos NãoTripulados Monitorização