Logo do repositório
 
Publicação

VIRIATO: Visual-Action Reinforcement Integrator for Actor-Critic with Temporal Observations

2026-04-22Comunicação em conferência Acesso aberto
Ver registo simples
datacite.subject.fosEngenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática
datacite.subject.sdg09:Indústria, Inovação e Infraestruturas
datacite.subject.sdg12:Produção e Consumo Sustentáveis
dc.contributor.authorCampanhã, João
dc.contributor.authorNeves, Francisco
dc.contributor.authorPinto, Andry
dc.contributor.authorBENEDITA CAMPOS NEVES MALHEIRO, MARIA
dc.contributor.editorFonseca, Pedro
dc.contributor.editorMoreira, António
dc.contributor.editorNeto, Pedro
dc.contributor.editorMorais, Pedro
dc.contributor.editorLima, José
dc.date.accessioned2026-07-10T13:44:19Z
dc.date.available2026-07-10T13:44:19Z
dc.date.issued2026-04-22
dc.description.abstractThe Visual-Action Reinforcement Integrator for Actor-Critic with Temporal Observations (VIRIATO) is a compact multi-input feature-extractor architecture designed to enable robust visual navigation of Unmanned Aerial Vehicles (UAVs) conducting close-range inspection of photovoltaic arrays. The target task of low-altitude flight over dynamic, visually variable surfaces without privileged information is inherently partially observable. VIRIATO augments stacked image observations with a short history of recent past actions, producing a richer latent state for the Soft Actor-Critic (SAC) agent. Training is performed with domain randomization to expose the policy to diverse lighting, backgrounds and panel layouts. In simulation, VIRIATO yields faster learning and improved sample efficiency compared to a standard image-only Convolutional Neural Network (CNN) feature extractor, achieving lower position and yaw errors and substantially better robustness under image perturbations while retaining high task completion rates. The architecture is intentionally simple and general: it improves temporal awareness without adding complex recurrence, and it could be adapted to other perception-driven robotic tasks. These results demonstrate that integrating historical action data with visual encoding, together with domain randomization, is an effective way to achieve reliable autonomous vision-based navigation.eng
dc.identifier.citationJ. Campanhã, F. Neves, B. Malheiro and A. Pinto, "VIRIATO: Visual-Action Reinforcement Integrator for Actor-Critic with Temporal Observations," 2026 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC), Barcelos, Portugal, 2026, pp. 214-221, doi: 10.1109/ICARSC70216.2026.11523320.
dc.identifier.doi10.1109/icarsc70216.2026.11523320
dc.identifier.isbn979-8-3195-1711-1
dc.identifier.urihttp://hdl.handle.net/10400.22/32569
dc.language.isoeng
dc.peerreviewedyes
dc.publisherIEEE
dc.relation.hasversionhttps://ieeexplore.ieee.org/document/11523320
dc.relation.ispartof2026 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC)
dc.relation.ispartofseriesIEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC)
dc.rights.uriN/A
dc.subjectReinforcement Learning (RL)
dc.subjectVisual navigation
dc.subjectUnmanned Aerial Vehicles (UAVs)
dc.subjectDomain Randomization
dc.subjectPhotovoltaic (PV) array inspection
dc.subjectSimulation
dc.titleVIRIATO: Visual-Action Reinforcement Integrator for Actor-Critic with Temporal Observationseng
dc.typeconference paper
dspace.entity.typePublication
oaire.citation.conferenceDate2026-04-23
oaire.citation.conferencePlaceBarcelos, Portugal
oaire.citation.endPage221
oaire.citation.startPage214
oaire.citation.title2026 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC)
oaire.versionhttp://purl.org/coar/version/c_ab4af688f83e57aa
person.familyNameBENEDITA CAMPOS NEVES MALHEIRO
person.givenNameMARIA
person.identifier.ciencia-id7A15-08FC-4430
person.identifier.orcid0000-0001-9083-4292
relation.isAuthorOfPublicationbabd4fda-654a-4b59-952d-6113eebbb308
relation.isAuthorOfPublication.latestForDiscoverybabd4fda-654a-4b59-952d-6113eebbb308

Ficheiros

Principais
A mostrar 1 - 1 de 1
Miniatura
Nome:
conference_101719.pdf
Tamanho:
12.73 MB
Formato:
Adobe Portable Document Format
Ver/Abrir
Licença
A mostrar 1 - 1 de 1
Miniatura indisponível
Nome:
license.txt
Tamanho:
4.03 KB
Formato:
Item-specific license agreed upon to submission
Descrição:
Ver/Abrir

Coleções

ISEP – LSA – Comunicações em eventos científicos
ISEP – DEE – Livro, parte de livro, ou capítulo de livro