Banca de DEFESA: RAFAELA OLIVEIRA REY

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
DISCENTE : RAFAELA OLIVEIRA REY
DATA : 30/06/2020
HORA: 09:00
LOCAL: PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA CIVIL
TÍTULO:

Computerized System for Safety Inspection in UAV Supported Construction Sites, Mobile Devices and Pattern Recognition Techniques

PALAVRAS-CHAVES:

Safety inspection. Construction site. Pattern recognition in image. UAS. UAV. Artificial Neural Networks.

PÁGINAS: 254
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Civil
RESUMO:

The construction sector is known for its high accident rates and neglect of working conditions. One of the main reasons that limit the quality of safety management decision making is access to information promptly. With the advent of industry 4.0, the way of carrying out the processes tends to change, making data collection, processing, and visualization operations more dynamic and automated. Despite the efforts for the insertion of digital technologies into construction projects, few studies have aimed at inspecting general safety conditions and inspecting collective protection equipment (CPE). Thus, this study aims to propose a computerized system to support the safety inspection on-site, focusing on the general conditions of the construction site and collective
protection equipment using UAS, mobile devices, and pattern recognition. The computerized system prototype was divided into two modules: (1) Inspection of general safety conditions, through the digital checklist based on Brazilian standards such as NR 18 and NR 35, and (2) Guardrail inspection, using artificial neural networks and color systems. The strategy adopted was Design Science Research. The system design included the definition of technological and technical information requirements for the development of the two modules. The implementation and evaluation of module 1 occurred at two construction sites, totalizing 55 inspections, and involving seven direct users. The tests of the algorithms developed for module 2 used 1535 images obtained with UAS of a guardrail prototype (with and without a marker), guardrails from the construction site, and 891 images from the GETEC and the internet database. The evaluations showed that module 1 of the computerized system could contribute to better communication between the project's stakeholders, allows the visualization of unsafe situations that were not previously evidenced and provides information for decision making with immediate corrections at the construction site. Module 2 showed 90% accuracy in the classification of the type of guardrails material, and errors of less than 1.0 cm in the dimensions obtained through the processing of the CPE images of the work, these results were favorable for its application in construction sites. Therefore, the
practical implication of this research involves improving the conditions of the work environment, through accurate and necessary information for a more conscious decision making. Besides, as a theoretical contribution, this research provides an understanding of the potentials and the limitations of the system application using UAS, in conjunction with mobile devices and pattern recognition techniques, to increase the efficiency of safety inspections at construction sites. This work also increases the understanding of possible changes from the adoption of these technologies in the Construction 4.0 context.

MEMBROS DA BANCA:
Presidente - 2618348 - DAYANA BASTOS COSTA
Externo à Instituição - Fabiano Rogerio Corre?a - USP
Externo à Instituição - Sheyla Mara Baptista Serra - UFSCAR