Detecção de buracos em estradas: uma abordagem automatizada baseada na transformada Wavelet de Haar

• Ano de defesa: 2019
• Autor(a) principal: Rodrigues, Ricardo Silveira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/26339/0013000005548
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Ciência da Computação; UFSM; Programa de Pós-Graduação em Ciência da Computação; Centro de Tecnologia
• Palavras-chave em Português: Buracos nas rodovias; Detecção de automática de buracos; Vibração; Acelerômetro; Wavelets; Automatic pothole detection; Vibration; Accelerometer; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/19334

Potholes in highways and roads cause various types of inconvenience to users such as passenger discomfort, damage to vehicle components and can cause accidents. In Brazil, the latest report by the National Transportation Confederation (CNT) in 2018 found that 50% of the 107,000 kilometers of paved highways surveyed in the country were classified as regular, bad or bad. Thus, general condition of roads is not good. Actually, this classification is done manually and is subject to errors or open to interpretations linked to the evaluators. Unfortunately, this situation is not very different in other countries. In India, the potholes are great cause of serious traffic accidents. Even the drivers of England are affected by the potholes in the roads. Thus, this work proposes an automated system for pothole detection using a wavelet approach in accelerometer signals. The use of accelerometers for pothole detection is not new. Typically, solutions that use accelerometers are based on thresholds, which indicate the existence of potholes, and are manually calibrated. The disadvantage of manual calibration is that it is dependent on an expert observer to perform an accurate adjustment. As a differential, in this work, the wavelet approach allows computing these thresholds automatically, without using manual calibration. Experiments were carried out in controlled and real environments, which confirm the efficiency of the proposed solution. This is an important step in the search for a full automation process for pothole detection in roads and highways.