Banca de QUALIFICAÇÃO: KAROLAYNE SANTOS DE AZEVEDO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : KAROLAYNE SANTOS DE AZEVEDO
DATE: 25/09/2024
TIME: 14:00
LOCAL: meet.google.com/ekk-rhmw-ary e no LANCE/nPITI
TITLE:

Identification of Driving Styles and Associated Risks in Semi-Autonomous Vehicles Using Artificial Intelligence Techniques

KEY WORDS:

Semi-Autonomous Vehicles, Artificial Intelligence, Ocular Entropy, Driving Styles, Driver Behavior.

PAGES: 70
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUMMARY:

Traditionally, drivers were responsible for the total control of the vehicle, but with the introduction of autonomous and semi-autonomous driving systems, this dynamic is being redefined. This work aims to identify and analyze driving style, behavior, and driver state using traditional and advanced artificial intelligence (AI) techniques from physiological data of 26 drivers who participated in the experiment. The research focuses on understanding how different driving styles and physiological conditions can influence interaction with autonomous and semi-autonomous systems. This work is part of an international cooperation involving partners in Brazil, with the participation of the Federal University of Rio Grande do Norte (UFRN) and the University of Granada (UGR). So far, eye movements have been used to analyze the driver's fatigue and distraction states, being among the most used neurobehavioral indices to track fatigue. From these data, ocular entropy images of the participants were generated, subsequently transformed into numerical vectors of length 1X768 using the embedding technique. These vectors were then applied in AI techniques to assess the results. Based on the ocular entropy data, the research obtained two main stages: prediction of the vehicle's manual and autonomous states, achieving 88.78% accuracy and 94.59% precision among five traditional machine learning (ML) models. The second stage of the research consists of identifying participants' driving styles into three groups (Consistent and Stable Driving, Aggressive Driving, and Aggressive and Variable Driving with Lane Changes and Sharp Acceleration) using clusters and radar charts. To obtain these results, the ocular entropy images were grouped into two groups, and then driving characteristics from the simulator were selected according to the respective group. These results were displayed in radar charts for each individual. Data related to the blink rate per minute were used to identify possible signs of fatigue in drivers during the manual driving state. The results reveal that most individuals showing signs of fatigue have a more aggressive driving style. Compared to the literature, no work uses only ocular entropy information as input to identify behaviors or vehicle states. This research faces significant challenges, including individual variability in driving styles and physiological responses of each experiment participant, as well as the need for precise and high-quality data.

COMMITTEE MEMBERS:
Presidente - 1837240 - MARCELO AUGUSTO COSTA FERNANDES
Interno - 1153006 - LUIZ AFFONSO HENDERSON GUEDES DE OLIVEIRA
Interna - ***.640.764-** - MARIANNE BATISTA DINIZ DA SILVA - UFRN
Externo ao Programa - 3083298 - RENAN CIPRIANO MOIOLI - UFRN