Estudo sobre metodologias de controle de vants para aplicação na criação de sistemas de inspeção

Abstract

This dissertation investigates visual control and positioning methodologies for unmanned aerial vehicles (UAVs) operating in environments without GNSS availability. It proposes an architecture that combines fixed cameras and fiducial markers distributed throughout the environment, enabling accurate pose estimation through computer vision. This configuration provides a low-cost, high-precision, and easily deployable solution, while centralizing processing and reducing the computational load on board the UAV. All experiments were conducted in an indoor environment with approximate dimensions of 5.5 m × 4.0 m × 3.0 m, although the proposed architecture presents no intrinsic limitations for outdoor use. External interferences, such as lighting variations, were not considered during testing. The control algorithms employed are manually tuned PID controllers, since the study focuses on the positioning architecture rather than controller optimization. Experimental results demonstrated average errors below 20 cm and performance comparable or superior to related visual positioning approaches reported in the literature. The proposed system contributes to the development of visual navigation solutions for UAVs in GNSS-denied scenarios, offering a precise, cost-effective, and reliable alternative suitable for inspection and monitoring tasks in confined or structured environments.