Learning Task-Agnostic Action Spaces for Movement Optimization

Simple item page
dc.contributorAalto-yliopistofi
dc.contributorAalto Universityen
dc.contributor.authorBabadi, Aminen_US
dc.contributor.authorVan de Panne, Michielen_US
dc.contributor.authorLiu, Carenen_US
dc.contributor.authorHamalainen, Perttuen_US
dc.contributor.departmentDepartment of Computer Scienceen
dc.contributor.departmentDepartment of Mediaen
dc.contributor.groupauthorProfessorship Hämäläinen Perttuen
dc.contributor.groupauthorComputer Science Professorsen
dc.contributor.groupauthorComputer Science - Visual Computing (VisualComputing)en
dc.contributor.organizationUniversity of British Columbiaen_US
dc.contributor.organizationStanford Universityen_US
dc.date.accessioned2022-12-14T10:16:40Z
dc.date.available2022-12-14T10:16:40Z
dc.date.issued2022-12en_US
dc.descriptionPublisher Copyright: CCBY
dc.description.abstractWe propose a novel method for exploring the dynamics of physically based animated characters, and learning a task-agnostic action space that makes movement optimization easier. Like several previous papers, we parameterize actions as target states, and learn a short-horizon goal-conditioned low-level control policy that drives the agent's state towards the targets. Our novel contribution is that with our exploration data, we are able to learn the low-level policy in a generic manner and without any reference movement data. Trained once for each agent or simulation environment, the policy improves the efficiency of optimizing both trajectories and high-level policies across multiple tasks and optimization algorithms. We also contribute novel visualizations that show how using target states as actions makes optimized trajectories more robust to disturbances; this manifests as wider optima that are easy to find. Due to its simplicity and generality, our proposed approach should provide a building block that can improve a large variety of movement optimization methods and applications.en
dc.description.versionPeer revieweden
dc.format.mimetypeapplication/pdfen_US
dc.identifier.citationBabadi, A, Van de Panne, M, Liu, C & Hamalainen, P 2022, ' Learning Task-Agnostic Action Spaces for Movement Optimization ', IEEE Transactions on Visualization and Computer Graphics, vol. 28, no. 12, pp. 4700-4712 . https://doi.org/10.1109/TVCG.2021.3100095en
dc.identifier.doi10.1109/TVCG.2021.3100095en_US
dc.identifier.issn1077-2626
dc.identifier.otherPURE UUID: 5e581a1e-4ab5-48af-8f74-ec3fce85cddaen_US
dc.identifier.otherPURE ITEMURL: https://research.aalto.fi/en/publications/5e581a1e-4ab5-48af-8f74-ec3fce85cddaen_US
dc.identifier.otherPURE LINK: http://www.scopus.com/inward/record.url?scp=85112599716&partnerID=8YFLogxKen_US
dc.identifier.otherPURE FILEURL: https://research.aalto.fi/files/94735096/Learning_Task_Agnostic_Action_Spaces_for_Movement_Optimization.pdfen_US
dc.identifier.urihttps://aaltodoc.aalto.fi/handle/123456789/118150
dc.identifier.urnURN:NBN:fi:aalto-202212146890
dc.language.isoenen
dc.publisherIEEE Computer Society
dc.relation.ispartofseriesIEEE Transactions on Visualization and Computer Graphicsen
dc.rightsopenAccessen
dc.subject.keywordaction spaceen_US
dc.subject.keywordAerospace electronicsen_US
dc.subject.keywordhierarchical reinforcement learningen_US
dc.subject.keywordmovement optimizationen_US
dc.subject.keywordOptimizationen_US
dc.subject.keywordpolicy optimizationen_US
dc.subject.keywordReinforcement learningen_US
dc.subject.keywordSplines (mathematics)en_US
dc.subject.keywordTask analysisen_US
dc.subject.keywordTrainingen_US
dc.subject.keywordtrajectory optimizationen_US
dc.subject.keywordTrajectory optimizationen_US
dc.titleLearning Task-Agnostic Action Spaces for Movement Optimizationen
dc.typeA1 Alkuperäisartikkeli tieteellisessä aikakauslehdessäfi
dc.type.versionpublishedVersion
Files
Collections
[article-cris] Perustieteiden korkeakoulu / SCI