Artigo em periódico
Autonomous Navigation for Mobile Robots in Unstructured Environments Using Deep Reinforcement Learning
Victor Figueiredo Lucena Junior, Raimundo Carlos Silvério Freire, Iury Valente de Bessa
Resumo
Resumo
This paper presents an end-to-end autonomous navigation system for mobile robots operating in unstructured environments using deep reinforcement learning. We develop a navigation policy based on Deep Q-Network (DQN) that learns directly from raw sensor data (RGB-D camera and LiDAR) without requiring explicit maps. The proposed approach is trained in simulation using domain randomization techniques and successfully transferred to real robots. Extensive experiments in challenging real-world scenarios demonstrate robust navigation performance, including obstacle avoidance, goal reaching, and recovery behaviors in cluttered environments typical of the Amazon region.
Autoria
Autores (3)
projeto relacionado
Projeto Relacionado
palavras-chave
Palavras-chave
Autonomous NavigationMobile RoboticsDeep Reinforcement LearningDQNUnstructured EnvironmentsObstacle Avoidance
notícias relacionadas
Notícias Relacionadas (1)
Volume
162
Páginas
104365
Qualis
A1
Fator de impacto
4.3
Quartil
Q1
Status
Published
links
Links e arquivos
Autonomous Navigation for Mobile Robots in Unstructured Environments Using Deep Reinforcement Learning
Citar publicação
Artigo em periódico