A data collecting framework for human reliability analysis via game engine based simulators

• Ano de defesa: 2018
• Autor(a) principal: ANDRADE, Marcos Vinicius Pinto de
• Orientador(a): MOURA, Márcio José das Chagas
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Pernambuco
• Programa de Pós-Graduação: Programa de Pos Graduacao em Engenharia de Producao
• País: Brasil
• Palavras-chave em Português: Engenharia de Produção; Simuladores; Realidade virtual; Motores de jogo; Confiabilidade Humana
• Link de acesso: https://repositorio.ufpe.br/handle/123456789/33596

The concern about human behavior and performance have become a major issue in almost every economic activity. Human errors have been in the top list for accident causes for years, as can be found in scientific literature and media, leading sometimes to huge death tolls, material losses, and environmental damages. The lack of good quality datasets for Human Reliability Assessment (HRA) studies has been an undesirable presence in the technological and academic areas, as also is the shortage of funding for research fostering nowadays in Brazil. The above-mentioned facts are unquestionable and can be easily found in scientific literature and in the media in general. The present work intends to address this situation by proposing an alternative approach to collecting HRA data from simulator sessions. The research also takes an approach under the perspective of small research teams, generated by the shortage of funding resources, and with that in mind analyses the use of Game Engines (GE) in the creation of Virtual Environments (VE) in 3D as a streamlined and more budget-conscious approach. Game Engines are one of the most advanced technological tools nowadays, combining 3D and 2D graphic engines, programming languages, physics simulation, web interactivity, business intelligence (BI) and many more, an achievement only possible thanks to the videogame industry, a multibillionaire business that grows each year. To validate the idea an experiment was conducted with a VE entirely created with a GE for the specific scenario of refinery plant evacuation under toxic cloud release, a scenario that fits well the technique. All good practices studied, tools developed, and assets created were condensed in the form of a framework for posterior use by the scientific community in general. Also, a Bayesian Belief Network (BBN) was created with the collected data validating the tool for HRA use. The study generated valuable insights into human behavior and generated good quality datasets.