Detecção e rastreamento de veículos em movimento para automóveis robóticos autônomos

• Ano de defesa: 2015
• Autor(a) principal: Amaral, Eduardo Max Amaro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Informática; Centro Tecnológico; UFES; Programa de Pós-Graduação em Informática
• Palavras-chave em Português: Navegação de robôs móveis; Veículos autônomos; Detectores; Movimento; Rastreamento automático; Robótica; Sensoriamento remoto; Ciência da Computação; 004
• Link de acesso: http://repositorio.ufes.br/handle/10/9863

In this work, it was investigated the problem of detection and tracking of moving objects (DATMO) for autonomous robotic vehicles. DATMO involves the detection of each moving object in the environment around the autonomous vehicle and its tracking, i.e., estimation of its state (e.g., position, orientation and velocity) over time. The autonomous vehicle needs to estimate the state of objects over time, so that it can predict their states a few seconds later for purposes of mapping, localization and navigation. It was proposed a DATMO system for the detection and tracking of multiple moving vehicles in the environment around an autonomous vehicle using a Light Detection and Ranging sensor (LIDAR) 3D. The proposed DATMO system operates in three steps: segmentation, association and tracking. At each sensor scan, in the segmentation step, the 3D points associated with the ground plane are removed; the 3D point cloud is segmented into clusters of points using the Euclidean distance, wherein each cluster represents an object in the environment; and the clusters related to curbs are removed. In association step, the objects observed in the current scan sensor are associated with the same objects observed in previous scans using the nearest neighbor algorithm. Finally, in the tracking step, the states of objects are estimated using a particle filter. Objects with velocity above a given threshold are considered moving vehicles. The performance of the proposed DATMO system was evaluated using data from a LIDAR 3D sensor, besides data from other sensors, collected by an autonomous vehicle along a ring road around the campus of the Federal University of Espírito Santo (Universidade Federal do Espírito Santo - UFES). The experimental results showed that the proposed DATMO system was able to detect and track with good performance multiple moving vehicles on the environment around the autonomous vehicle.