| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 70.25 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Devido a esta utilização mais alargada dos UAV’s, esta dissertação pretende analisar a possibilidade de utilização de mapeamento 3D e odometria baseado apenas em LiDAR como forma de substituir outros sensores, como o GPS em caso de falha de conexão ou jamming do mesmo. De forma a responder a esta questão, esta dissertação irá analisar a possibilidade de utilização de implementações previamente desenvolvidas e baseadas em LOAM, até então testadas em veículos terrestres. Para além de testar esta possibilidade, pretende ainda identificar a implementação indicada para utilizar no contexto pretendido, uma vez que a sua utilização num UAV com processamento em tempo real apresenta maiores restrições do que as impostas aos veículos terrestres. Estas restrições advêm principalmente da reduzida versatilidade na adição de capacidade de processamento. De forma a realizar a análise, foram assim realizados dois datasets que permitiram perceber as restrições e cuidados a ter com o plano de voo caso o UAV esteja equipado com uma implementação baseada em LOAM. Estes testes permitiram ainda compreender qual das implementações seria a mais indicada para uma possível implementação num UAV.
Due to this wider use of UAV’s, this dissertation aims to analyze the possibility of using 3D mapping and odometry based only on LiDAR as a way to replace other sensors, such as GPS in case of connection failure or jamming of the signal. In order to answer this question, this dissertation will analyze the possibility of using implementations previously developed and based on LOAM, hitherto tested in land vehicles. In addition to testing this possibility, it also intends to identify the appropriate implementation to use in the intended context, since its use in an UAV with real-time processing presents greater restrictions than those imposed on land vehicles. These restrictions stem mainly from the reduced versatility in adding processing power. In order to do the analyses, two datasets where made what enabled us to understand the constraints to look into when planning a flight plan for an UAV equipped with one of the LOAM based implementations. This tests, enabled us to also understand wich of the implementations would the the most suitable to apply on an UAV.
Due to this wider use of UAV’s, this dissertation aims to analyze the possibility of using 3D mapping and odometry based only on LiDAR as a way to replace other sensors, such as GPS in case of connection failure or jamming of the signal. In order to answer this question, this dissertation will analyze the possibility of using implementations previously developed and based on LOAM, hitherto tested in land vehicles. In addition to testing this possibility, it also intends to identify the appropriate implementation to use in the intended context, since its use in an UAV with real-time processing presents greater restrictions than those imposed on land vehicles. These restrictions stem mainly from the reduced versatility in adding processing power. In order to do the analyses, two datasets where made what enabled us to understand the constraints to look into when planning a flight plan for an UAV equipped with one of the LOAM based implementations. This tests, enabled us to also understand wich of the implementations would the the most suitable to apply on an UAV.
Description
Keywords
UAV LiDAR Odometria Mapeamento Benchmark Odometry Mapping