Desenvolvimento de plataforma robótica com sistema de navegação autônoma para aplicação na agricultura de precisão
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Federal de Viçosa
Abstract
A Agricultura de Precisão é caracterizada pela otimização do manejo e insumos agrícolas, utilizando sensores, programas de computador e máquinas agrícolas automatizadas. O desenvolvimento de robôs agrícolas autônomos, com tecnologias como Sistema Global de Navegação por Satélite com Cinemática em Tempo Real (GNSS-RTK), enfrenta desafios de custo e integração eficiente de Unidade de Medição Inercial (IMU) para precisão na navegação. A utilização de componentes de baixo custo possibilita fomentar a automação na agricultura de precisão e aumentar a lucratividade na produção agrícola. O presente trabalho teve como objetivo desenvolver uma plataforma robótica de baixo custo com navegação autônoma. Para orientação e navegação utilizou-se módulos de GNSS-RTK e sensor de IMU. Desenvolveu-se um algoritmo de navegação baseado em cinemática de veículo diferencial, associado a um controlador Proporcional e Integral (PI) de direção, que, dada uma rota, seguia ponto alvo a ponto alvo até concluir o percurso. Foram avaliadas duas formas de desenvolvimento do mapa da rota planejada: a primeira distribuiu-se os pontos alvo apenas nos vértices da rota planejada, e a segunda os pontos alvo foram alocados a cada três metros. Foram coletados dados de erro lateral por meio do módulo GNSS-RTK e por uma webcam. Com os dados obtidos do GNSS- RTK avaliou-se o desempenho do algoritmo de controle da plataforma robótica em seguir uma rota pré-definida. Com os dados extraídos das imagens foi analisado o erro global do sistema, já que se conhecia a posição real do robô em relação à rota pré-definida. A principal métrica de desempenho avaliado foi o percentual da rota que a plataforma robótica percorreu além de uma faixa de 0,3 m para ambos os lados em relação à rota pré-definida, situação essa que ocorreria virtualmente uma colisão com uma planta de café. Obteve-se os seguintes resultados considerando o melhor e pior resultado dos diferentes ensaios: maior erro lateral: 0,233 m à 1,525 m, média do erro absoluto: 0,007 m à 0,224 m, desvio padrão: 0,019 m à 0,225 m erro quadrático médio: 0,043 m à 0,314 m. Mostrou-se viável a utilização de componentes de baixo custo no desenvolvimento de uma plataforma robótica. O algoritmo desenvolvido controlou o movimento da plataforma em uma plantação de café simulada, destacando-se na navegação retilínea, manobras e retorno ao ponto de origem. Ao comparar duas rotas diferentes, uma com pontos alvo nos vértices e outra a cada 3 metros, a segunda resultou em menores erros laterais. A escolha da estratégia de geração de rota revelou-se crucial para a precisão da navegação autônoma. Palavras-chave: Robôs Agrícolas. GNSS-RTK. IMU. Componentes de Baixo Custo. Algoritmo de Navegação.
Precision Agriculture is characterized by optimizing the management and agricultural inputs, using sensors, computer programs, and automated agricultural machinery. The development of autonomous agricultural robots, employing technologies like Global Navigation Satellite System with Real-Time Kinematics (GNSS-RTK), faces challenges of cost and efficient integration of Inertial Measurement Unit (IMU) for precise navigation. The use of low-cost components enables the promotion of automation in precision agriculture, thereby increasing profitability in agricultural production. The present work aimed to develop a low-cost robotic platform with autonomous navigation. For guidance and navigation, GNSS-RTK modules and an IMU sensor were utilized. A navigation algorithm based on differential vehicle kinematics was developed, associated with a Proportional and Integral (PI) steering controller, which, given a route, followed target point to target point until completing the route. Two ways of developing the map of the planned route were evaluated: the first distributed target points only at the vertices of the planned route, and the second the target points were allocated every three meters. Lateral error data were collected through the GNSS-RTK module and a webcam. Using the GNSS-RTK data, the performance of the robotic platform control algorithm in following a pre-defined route was assessed. Images data were analyzed to determine the overall system error, considering the known real position of the robot in relation to the predefined route. The primary performance metric evaluated was the percentage of the route that the robotic platform traversed beyond a 0.3 m range on both sides of the predefined route, a situation that would virtually result in a collision with a coffee plant. The following results were obtained considering the best and worst results of the different trials: maximum lateral error: 0.233 m to 1.525 m, average absolute error: 0.007 m to 0.224 m, standard deviation: 0.019 m to 0.225 m, mean square error: 0.043 m to 0.314 m. The use of low- cost components in the development of a robotic platform was proven to be viable. The developed algorithm controlled the platform's movement in a simulated coffee plantation, excelling in straight-line navigation, maneuvers, and return to the starting point. When comparing two different routes, one with target points at the vertices and the other every 3 meters, the second had lower lateral errors. The choice of route generation strategy proved to be crucial for the accuracy of autonomous navigation. Keywords: Agricultural Robots. GNSS-RTK. IMU. Low-Cost Components. Navigation Algorithm.
Precision Agriculture is characterized by optimizing the management and agricultural inputs, using sensors, computer programs, and automated agricultural machinery. The development of autonomous agricultural robots, employing technologies like Global Navigation Satellite System with Real-Time Kinematics (GNSS-RTK), faces challenges of cost and efficient integration of Inertial Measurement Unit (IMU) for precise navigation. The use of low-cost components enables the promotion of automation in precision agriculture, thereby increasing profitability in agricultural production. The present work aimed to develop a low-cost robotic platform with autonomous navigation. For guidance and navigation, GNSS-RTK modules and an IMU sensor were utilized. A navigation algorithm based on differential vehicle kinematics was developed, associated with a Proportional and Integral (PI) steering controller, which, given a route, followed target point to target point until completing the route. Two ways of developing the map of the planned route were evaluated: the first distributed target points only at the vertices of the planned route, and the second the target points were allocated every three meters. Lateral error data were collected through the GNSS-RTK module and a webcam. Using the GNSS-RTK data, the performance of the robotic platform control algorithm in following a pre-defined route was assessed. Images data were analyzed to determine the overall system error, considering the known real position of the robot in relation to the predefined route. The primary performance metric evaluated was the percentage of the route that the robotic platform traversed beyond a 0.3 m range on both sides of the predefined route, a situation that would virtually result in a collision with a coffee plant. The following results were obtained considering the best and worst results of the different trials: maximum lateral error: 0.233 m to 1.525 m, average absolute error: 0.007 m to 0.224 m, standard deviation: 0.019 m to 0.225 m, mean square error: 0.043 m to 0.314 m. The use of low- cost components in the development of a robotic platform was proven to be viable. The developed algorithm controlled the platform's movement in a simulated coffee plantation, excelling in straight-line navigation, maneuvers, and return to the starting point. When comparing two different routes, one with target points at the vertices and the other every 3 meters, the second had lower lateral errors. The choice of route generation strategy proved to be crucial for the accuracy of autonomous navigation. Keywords: Agricultural Robots. GNSS-RTK. IMU. Low-Cost Components. Navigation Algorithm.
Description
Citation
BALTAZAR, Jamil de Almeida. Desenvolvimento de plataforma robótica com sistema de navegação autônoma para aplicação na agricultura de precisão. 2024. 58 f. Dissertação (Mestrado em Engenharia Agrícola) - Universidade Federal de Viçosa, Viçosa. 2024.
