| Texto completo | |
| Autor(es): |
Rossi, Leonardo L.
;
Berto, Leticia
;
Costa, Paula P.
;
Gudwin, Ricardo
;
Colombini, Esther
;
Simoes, Alexandre
Número total de Autores: 6
|
| Tipo de documento: | Artigo Científico |
| Fonte: | COGNITIVE SYSTEMS RESEARCH; v. 95, p. 16-pg., 2026-01-01. |
| Resumo | |
Reinforcement learning (RL) methods inspired by cognitive architectures are crucial for empowering autonomous agents to tackle complex, dynamic tasks. This study evaluates two RL-based drive optimization strategies-1-LDO and 2-LDO-within the framework of cognitive architectures for autonomous robots. 1-LDO integrates both motivational drives into a single learning model, whereas 2-LDO separates them into distinct models, allowing for modular learning. Grounded in Hull's Drive Theory, we explore early versus late selection mechanisms to optimize drive reduction through RL, particularly in agents driven by curiosity and survival imperatives. Through reward and stress analyses, we demonstrate that Deep Q-Network (DQN) agents outperform traditional Q-Learning approaches in fine-grained environments, with the 2-LDO configuration showing marked advantages due to its modular design. In contrast, in coarser environments, 2-LDO combined with Q-Learning achieves superior efficiency, offering faster drive regulation at reduced computational cost. These results suggest that early selection mechanisms, aligned with Hull's theoretical principles, may provide the most effective strategy for optimizing drive-based behaviors in autonomous agents. (AU) | |
| Processo FAPESP: | 20/09850-0 - Centro de Pesquisa Aplicada em Inteligência Artificial: impulsionando a transformação das indústrias rumo ao padrão 5.0 |
| Beneficiário: | Jefferson de Oliveira Gomes |
| Modalidade de apoio: | Auxílio à Pesquisa - Programa Centros de Pesquisa Aplicada |
| Processo FAPESP: | 13/07559-3 - Instituto Brasileiro de Neurociência e Neurotecnologia - BRAINN |
| Beneficiário: | Fernando Cendes |
| Modalidade de apoio: | Auxílio à Pesquisa - Centros de Pesquisa, Inovação e Difusão - CEPIDs |
| Processo FAPESP: | 21/07050-0 - Uma estrutura de aprendizagem incremental baseada em motivação para robótica |
| Beneficiário: | Letícia Mara Berto |
| Modalidade de apoio: | Bolsas no Brasil - Doutorado |