Reinforcement learning in autonomous vehicles with limited input

The automotive industry is currently undergoing two simultaneous revolutions: electrification and autonomy. In recent years, various regulations that apply to all manufacturers have mandated the incorporation of advanced driver assistance systems (ADAS) into their vehicles. Manufacturers are already...

ver descrição completa

Detalhes bibliográficos
Autor: Villalba Rodríguez, Raül
Formato: tesis de maestría
Fecha de publicación:2024
País:España
Recursos:Universitat Oberta de Catalunya (UOC)
Repositorio:O2, repositorio institucional de la UOC
OAI Identifier:oai:openaccess.uoc.edu:10609/151204
Acesso em linha:http://hdl.handle.net/10609/151204
Access Level:acceso abierto
Palavra-chave:automotive
autonomous driving
advanced driver assistance systems (ADAS)
level L5 autonomy
reinforcement learning
HD maps
Automated vehicles -- FMDP
Vehicles autònoms -- TFM
Descrição
Resumo:The automotive industry is currently undergoing two simultaneous revolutions: electrification and autonomy. In recent years, various regulations that apply to all manufacturers have mandated the incorporation of advanced driver assistance systems (ADAS) into their vehicles. Manufacturers are already delving into the development of Level L3 autonomous functions that can be homologated. Additionally, numerous institutions and companies are pushing the boundaries, aiming to create prototypes that achieve Level L5 autonomy, enabling fully autonomous driving. Within the scope of this project, we will delve into the cutting-edge developments in this field. Our objective is to implement and train a model utilizing reinforcement learning, and subsequently, compare its performance with other models. To ensure safety throughout the process, the training will be carried out within simulated environments. However, our ultimate goal is to seamlessly integrate the trained model into a real-world vehicle.