Matches in SemOpenAlex for { <https://semopenalex.org/work/W4225873172> ?p ?o ?g. }
- W4225873172 endingPage "4133" @default.
- W4225873172 startingPage "4126" @default.
- W4225873172 abstract "We tackle real-world long-horizon robot manipulation tasks through skill discovery. We present a bottom-up approach to learning a library of reusable skills from unsegmented demonstrations and use these skills to synthesize prolonged robot behaviors. Our method starts with constructing a hierarchical task structure from each demonstration through agglomerative clustering. From the task structures of multi-task demonstrations, we identify skills based on the recurring patterns and train goal-conditioned sensorimotor policies with hierarchical imitation learning. Finally, we train a meta controller to compose these skills to solve long-horizon manipulation tasks. The entire model can be trained on a small set of human demonstrations collected within 30 minutes without further annotations, making it amendable to real-world deployment. We systematically evaluated our method in simulation environments and on a real robot. Our method has shown superior performance over state-of-the-art imitation learning methods in multi-stage manipulation tasks. Furthermore, skills discovered from multi-task demonstrations boost the average task success by 8% compared to those discovered from individual tasks." @default.
- W4225873172 created "2022-05-05" @default.
- W4225873172 creator A5001594330 @default.
- W4225873172 creator A5030826237 @default.
- W4225873172 creator A5035249197 @default.
- W4225873172 date "2022-04-01" @default.
- W4225873172 modified "2023-10-18" @default.
- W4225873172 title "Bottom-Up Skill Discovery From Unsegmented Demonstrations for Long-Horizon Robot Manipulation" @default.
- W4225873172 cites W1529739011 @default.
- W4225873172 cites W1837533429 @default.
- W4225873172 cites W1907587592 @default.
- W4225873172 cites W2022760091 @default.
- W4225873172 cites W2027972120 @default.
- W4225873172 cites W2070683816 @default.
- W4225873172 cites W2097856935 @default.
- W4225873172 cites W2109910161 @default.
- W4225873172 cites W2112474089 @default.
- W4225873172 cites W2116064496 @default.
- W4225873172 cites W2132914434 @default.
- W4225873172 cites W2135639338 @default.
- W4225873172 cites W2307770531 @default.
- W4225873172 cites W2397253692 @default.
- W4225873172 cites W2892211305 @default.
- W4225873172 cites W2919658942 @default.
- W4225873172 cites W2963669336 @default.
- W4225873172 cites W2963927307 @default.
- W4225873172 cites W2964055695 @default.
- W4225873172 cites W2968344028 @default.
- W4225873172 cites W2982673782 @default.
- W4225873172 cites W3012366945 @default.
- W4225873172 cites W3012573144 @default.
- W4225873172 cites W3038298277 @default.
- W4225873172 cites W3039055967 @default.
- W4225873172 cites W3090369311 @default.
- W4225873172 cites W3128894241 @default.
- W4225873172 cites W3150915266 @default.
- W4225873172 cites W3174700686 @default.
- W4225873172 cites W3207612633 @default.
- W4225873172 cites W377632744 @default.
- W4225873172 cites W4246451932 @default.
- W4225873172 doi "https://doi.org/10.1109/lra.2022.3146589" @default.
- W4225873172 hasPublicationYear "2022" @default.
- W4225873172 type Work @default.
- W4225873172 citedByCount "6" @default.
- W4225873172 countsByYear W42258731722022 @default.
- W4225873172 countsByYear W42258731722023 @default.
- W4225873172 crossrefType "journal-article" @default.
- W4225873172 hasAuthorship W4225873172A5001594330 @default.
- W4225873172 hasAuthorship W4225873172A5030826237 @default.
- W4225873172 hasAuthorship W4225873172A5035249197 @default.
- W4225873172 hasBestOaLocation W42258731722 @default.
- W4225873172 hasConcept C105339364 @default.
- W4225873172 hasConcept C107457646 @default.
- W4225873172 hasConcept C111919701 @default.
- W4225873172 hasConcept C119857082 @default.
- W4225873172 hasConcept C126388530 @default.
- W4225873172 hasConcept C127413603 @default.
- W4225873172 hasConcept C132758656 @default.
- W4225873172 hasConcept C154945302 @default.
- W4225873172 hasConcept C15744967 @default.
- W4225873172 hasConcept C162324750 @default.
- W4225873172 hasConcept C177264268 @default.
- W4225873172 hasConcept C199360897 @default.
- W4225873172 hasConcept C201995342 @default.
- W4225873172 hasConcept C2780451532 @default.
- W4225873172 hasConcept C41008148 @default.
- W4225873172 hasConcept C50522688 @default.
- W4225873172 hasConcept C73555534 @default.
- W4225873172 hasConcept C77805123 @default.
- W4225873172 hasConcept C90509273 @default.
- W4225873172 hasConceptScore W4225873172C105339364 @default.
- W4225873172 hasConceptScore W4225873172C107457646 @default.
- W4225873172 hasConceptScore W4225873172C111919701 @default.
- W4225873172 hasConceptScore W4225873172C119857082 @default.
- W4225873172 hasConceptScore W4225873172C126388530 @default.
- W4225873172 hasConceptScore W4225873172C127413603 @default.
- W4225873172 hasConceptScore W4225873172C132758656 @default.
- W4225873172 hasConceptScore W4225873172C154945302 @default.
- W4225873172 hasConceptScore W4225873172C15744967 @default.
- W4225873172 hasConceptScore W4225873172C162324750 @default.
- W4225873172 hasConceptScore W4225873172C177264268 @default.
- W4225873172 hasConceptScore W4225873172C199360897 @default.
- W4225873172 hasConceptScore W4225873172C201995342 @default.
- W4225873172 hasConceptScore W4225873172C2780451532 @default.
- W4225873172 hasConceptScore W4225873172C41008148 @default.
- W4225873172 hasConceptScore W4225873172C50522688 @default.
- W4225873172 hasConceptScore W4225873172C73555534 @default.
- W4225873172 hasConceptScore W4225873172C77805123 @default.
- W4225873172 hasConceptScore W4225873172C90509273 @default.
- W4225873172 hasFunder F4320306994 @default.
- W4225873172 hasFunder F4320313876 @default.
- W4225873172 hasFunder F4320332180 @default.
- W4225873172 hasIssue "2" @default.
- W4225873172 hasLocation W42258731721 @default.
- W4225873172 hasLocation W42258731722 @default.
- W4225873172 hasOpenAccess W4225873172 @default.
- W4225873172 hasPrimaryLocation W42258731721 @default.
- W4225873172 hasRelatedWork W2007682987 @default.