Matches in SemOpenAlex for { <https://semopenalex.org/work/W4328007230> ?p ?o ?g. }
- W4328007230 endingPage "11" @default.
- W4328007230 startingPage "1" @default.
- W4328007230 abstract "The emerging embodied AI paradigm enables intelligent robots to learn like humans from interaction, and is thus considered an effective way approach to general artificial intelligence. Unfortunately, even the best performing agents still overfit and generalize poorly to unseen scenes, due to the limited scenes provided by embodied AI simulators. To alleviate this issue, we propose a scene augmentation strategy to scale up the scene diversity for interactive tasks and make the interactions more like the real world. Compared to existing methods focusing on improving diversity in observation space, our approach aims to automatically derive a new distribution of scene layout or object states to provide sufficient conditional transfer models for the agent to learn environmental invariant and irrelevant features through interaction. Specifically, we provide four representative and systematical scene augmentation methods that can derive scene variants for entities from different levels of a scene graph. We demonstrate the efficiency of our methods in the popular embodied AI simulator iGibson. To verify the effectiveness for interactive agents, we also conduct two representative interactive tasks with a proposed continuous action parameterized method. The evaluation results show that our scene augmentation strategy can boost the performance of interactive agents and generalize well to unseen scenes. Finally, we present a systematic generalization analysis using the proposed methods to explicitly estimate the ability of agents to generalize to new layouts, new objects, and new object states. We claim that the proposed methods are not limited to iGibson and can be extended to other interactive simulators. The code and additional information are available at: https://github.com/sanghongrui/SceneAug." @default.
- W4328007230 created "2023-03-22" @default.
- W4328007230 creator A5000024164 @default.
- W4328007230 creator A5006874566 @default.
- W4328007230 creator A5018569324 @default.
- W4328007230 creator A5031784349 @default.
- W4328007230 creator A5032983330 @default.
- W4328007230 date "2023-01-01" @default.
- W4328007230 modified "2023-10-15" @default.
- W4328007230 title "Scene Augmentation Methods for Interactive Embodied AI Tasks" @default.
- W4328007230 cites W1424654272 @default.
- W4328007230 cites W2103104224 @default.
- W4328007230 cites W2108598243 @default.
- W4328007230 cites W2157331557 @default.
- W4328007230 cites W2158782408 @default.
- W4328007230 cites W2167340365 @default.
- W4328007230 cites W2194775991 @default.
- W4328007230 cites W2605102758 @default.
- W4328007230 cites W2954996726 @default.
- W4328007230 cites W2962887844 @default.
- W4328007230 cites W2963748441 @default.
- W4328007230 cites W2990747716 @default.
- W4328007230 cites W2998508940 @default.
- W4328007230 cites W3039925688 @default.
- W4328007230 cites W3108332675 @default.
- W4328007230 cites W3137240333 @default.
- W4328007230 cites W3173760279 @default.
- W4328007230 cites W3174804976 @default.
- W4328007230 cites W3184282957 @default.
- W4328007230 cites W3192162805 @default.
- W4328007230 cites W3201915713 @default.
- W4328007230 cites W4200523064 @default.
- W4328007230 cites W4205900130 @default.
- W4328007230 cites W4207072548 @default.
- W4328007230 cites W4226174994 @default.
- W4328007230 cites W4283080798 @default.
- W4328007230 cites W4285234731 @default.
- W4328007230 doi "https://doi.org/10.1109/tim.2023.3259033" @default.
- W4328007230 hasPublicationYear "2023" @default.
- W4328007230 type Work @default.
- W4328007230 citedByCount "0" @default.
- W4328007230 crossrefType "journal-article" @default.
- W4328007230 hasAuthorship W4328007230A5000024164 @default.
- W4328007230 hasAuthorship W4328007230A5006874566 @default.
- W4328007230 hasAuthorship W4328007230A5018569324 @default.
- W4328007230 hasAuthorship W4328007230A5031784349 @default.
- W4328007230 hasAuthorship W4328007230A5032983330 @default.
- W4328007230 hasConcept C100609095 @default.
- W4328007230 hasConcept C103683099 @default.
- W4328007230 hasConcept C107457646 @default.
- W4328007230 hasConcept C119857082 @default.
- W4328007230 hasConcept C134306372 @default.
- W4328007230 hasConcept C154945302 @default.
- W4328007230 hasConcept C177148314 @default.
- W4328007230 hasConcept C179372163 @default.
- W4328007230 hasConcept C205711294 @default.
- W4328007230 hasConcept C22019652 @default.
- W4328007230 hasConcept C2781238097 @default.
- W4328007230 hasConcept C33923547 @default.
- W4328007230 hasConcept C41008148 @default.
- W4328007230 hasConcept C50644808 @default.
- W4328007230 hasConcept C90509273 @default.
- W4328007230 hasConceptScore W4328007230C100609095 @default.
- W4328007230 hasConceptScore W4328007230C103683099 @default.
- W4328007230 hasConceptScore W4328007230C107457646 @default.
- W4328007230 hasConceptScore W4328007230C119857082 @default.
- W4328007230 hasConceptScore W4328007230C134306372 @default.
- W4328007230 hasConceptScore W4328007230C154945302 @default.
- W4328007230 hasConceptScore W4328007230C177148314 @default.
- W4328007230 hasConceptScore W4328007230C179372163 @default.
- W4328007230 hasConceptScore W4328007230C205711294 @default.
- W4328007230 hasConceptScore W4328007230C22019652 @default.
- W4328007230 hasConceptScore W4328007230C2781238097 @default.
- W4328007230 hasConceptScore W4328007230C33923547 @default.
- W4328007230 hasConceptScore W4328007230C41008148 @default.
- W4328007230 hasConceptScore W4328007230C50644808 @default.
- W4328007230 hasConceptScore W4328007230C90509273 @default.
- W4328007230 hasFunder F4320321001 @default.
- W4328007230 hasFunder F4320321885 @default.
- W4328007230 hasFunder F4320335777 @default.
- W4328007230 hasFunder F4320335787 @default.
- W4328007230 hasFunder F4320335791 @default.
- W4328007230 hasLocation W43280072301 @default.
- W4328007230 hasOpenAccess W4328007230 @default.
- W4328007230 hasPrimaryLocation W43280072301 @default.
- W4328007230 hasRelatedWork W1522117956 @default.
- W4328007230 hasRelatedWork W1596535966 @default.
- W4328007230 hasRelatedWork W2023938433 @default.
- W4328007230 hasRelatedWork W2068486122 @default.
- W4328007230 hasRelatedWork W2942109448 @default.
- W4328007230 hasRelatedWork W2989932438 @default.
- W4328007230 hasRelatedWork W3011996705 @default.
- W4328007230 hasRelatedWork W3099765033 @default.
- W4328007230 hasRelatedWork W3103741966 @default.
- W4328007230 hasRelatedWork W3206529776 @default.
- W4328007230 hasVolume "72" @default.
- W4328007230 isParatext "false" @default.
- W4328007230 isRetracted "false" @default.