Implementation of Artificial Intelligence models for managing abnormal situations and emergencies in complex systems

• Grant number: 25/03749-0
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: May 01, 2025
• End date: April 30, 2029
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Carlos Henrique Costa Ribeiro
• Grantee: Emili Silva Bezerra
• Host Institution: Pró-Reitoria de Pós-Graduação e Pesquisa. Instituto Tecnológico de Aeronáutica (ITA). Ministério da Defesa (Brasil). São José dos Campos , SP, Brazil
• Company: Ministério da Defesa (Brasil). Instituto Tecnológico de Aeronáutica (ITA). Divisão de Engenharia Mecânica (IEM)
• Associated research grant: 21/11258-5 - Engineering Research Center for the Aerial Mobility of the Future (ERC-AMF), AP.PCPE

Abstract

Future aviation systems will include autonomous aircrafts and reduced crew operations. In both situations the introduction of better emergency managements techniques is desired, either for autonomous operation, or for aiding a potentially overwhelmed single pilot. Automation of current checklists may be performed, but this solution is not resilient to multiple failures: in those cases, pilot experience plays a major role in adapting and combining checklists in unforeseen strategies to cope with specific situations. More than automating protocols, autonomous decision making based on learning data is required, but due to the high reliability of aircraft systems, real data for training artificial intelligence (AI) models is not readily available for being considered of low relevance, thus synthetic data needs to be used if an autonomous solution is to be considered. This research intends to analyze and use state-of-the-art techniques to emulate an experienced pilot decision making, in order to analyze and deal with complex system failures and reconfigure the aircraft in unforeseen situations, using AI techniques over massive and possibly partially synthesized datasets.