| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 28.31 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A utilização de veículos aéreos não tripulados é recorrente nos dias de hoje em missões de vigilância, inspeção ou apoio a equipas de socorro em missões de busca e salvamento.
De forma a minimizar o erro humano e os custos de operação, estes sistemas são operados em modo autónomo, incluindo a fase da aterragem. A manobra de aterragem necessita de ser desempenhada com o menor erro de posição possível, pelo que as soluções clássicas através de recetores Global Navigation Satellite System (GNSS) em modo single, podem não ser suficientemente precisas.
Na dissertação propõe-se endereçar o desenvolvimento de um método de localização relativa que permita a aterragem autónoma de um Vertical Take-Off and Landing (VTOL) numa pista de aterragem móvel. Pretende-se a integração de uma baseline móvel Real Time Kinematic (RTK) entre a pista de aterragem e o VTOL, e combinar a baseline com a estimação de posição e atitude dada pelo sistema de visão a bordo.
O método desenvolvido de localização através de visão computacional consiste na deteção de um marcador visual ativo modelado para disparar sincronizado com a aquisição de imagens.
A qualidade do posicionamento e atitude fornecidos pela visão computacional em comparação com o sistema RTK e informação inercial, que serviram como ground-truth, foi aferida com recurso a testes experimentais.
The use of unmanned aerial vehicles is recurrent today in missions of surveillance, inspection or support to rescue teams in search and rescue missions. In order to minimize human error and operating costs, these systems are operated in an autonomous mode, including the landing maneuver. The landing maneuver needs to be performed with as little position error as possible, so classical solutions through single-mode GNSS receivers may not be sufficiently accurate. The present document proposes to address the development of a relative localization method allowing the autonomous landing of a VTOL on a mobile landing strip. The intention is to integrate an RTK mobile baseline between the landing pad and the VTOL, and combine the baseline with the position and attitude estimation given by the on-board vision system. The developed method of location through computer vision consists of the detection of an active visual marker, which is modelled to trigger synchronized with the acquisition of images. The quality of the positionion and attitude provided by the computer vision in comparison with the RTK system and inertial information, which served as ground-truth, was measured using experimental tests.
The use of unmanned aerial vehicles is recurrent today in missions of surveillance, inspection or support to rescue teams in search and rescue missions. In order to minimize human error and operating costs, these systems are operated in an autonomous mode, including the landing maneuver. The landing maneuver needs to be performed with as little position error as possible, so classical solutions through single-mode GNSS receivers may not be sufficiently accurate. The present document proposes to address the development of a relative localization method allowing the autonomous landing of a VTOL on a mobile landing strip. The intention is to integrate an RTK mobile baseline between the landing pad and the VTOL, and combine the baseline with the position and attitude estimation given by the on-board vision system. The developed method of location through computer vision consists of the detection of an active visual marker, which is modelled to trigger synchronized with the acquisition of images. The quality of the positionion and attitude provided by the computer vision in comparison with the RTK system and inertial information, which served as ground-truth, was measured using experimental tests.
Description
Keywords
Aterragem Autónoma Posicionamento Relativo UAV GNSS RTK Visão Computacional IMU Sincronismo Autonomous Landing Relative Postioning Computer Vision Synchronism