Matches in SemOpenAlex for { <https://semopenalex.org/work/W3096593047> ?p ?o ?g. }
Showing items 1 to 99 of
99
with 100 items per page.
- W3096593047 abstract "Transferring reinforcement learning policies trained in physics simulation to the real hardware remains a challenge, known as the sim-to-real gap. Domain randomization is a simple yet effective technique to address dynamics discrepancies across source and target domains, but its success generally depends on heuristics and trial-and-error. In this work we investigate the impact of randomized parameter selection on policy transferability across different types of domain discrepancies. Contrary to common practice in which kinematic parameters are carefully measured while dynamic parameters are randomized, we found that virtually randomizing kinematic parameters (e.g., link lengths) during training in simulation generally outperforms dynamic randomization. Based on this finding, we introduce a new domain adaptation algorithm that utilizes simulated kinematic parameters variation. Our algorithm, Multi-Policy Bayesian Optimization, trains an ensemble of universal policies conditioned on virtual kinematic parameters and efficiently adapts to the target environment using a limited number of target domain rollouts. We showcase our findings on a simulated quadruped robot in five different target environments covering different aspects of domain discrepancies." @default.
- W3096593047 created "2020-11-09" @default.
- W3096593047 creator A5015524477 @default.
- W3096593047 creator A5020531381 @default.
- W3096593047 creator A5036766747 @default.
- W3096593047 creator A5084848670 @default.
- W3096593047 date "2020-11-03" @default.
- W3096593047 modified "2023-09-27" @default.
- W3096593047 title "Policy Transfer via Kinematic Domain Randomization and Adaptation" @default.
- W3096593047 cites W1746819321 @default.
- W3096593047 cites W1979328769 @default.
- W3096593047 cites W2121863487 @default.
- W3096593047 cites W2477400028 @default.
- W3096593047 cites W2595845486 @default.
- W3096593047 cites W2602963933 @default.
- W3096593047 cites W2605102758 @default.
- W3096593047 cites W2736601468 @default.
- W3096593047 cites W2889970038 @default.
- W3096593047 cites W2911087563 @default.
- W3096593047 cites W2919616510 @default.
- W3096593047 cites W2963184939 @default.
- W3096593047 cites W2963859851 @default.
- W3096593047 cites W2968116426 @default.
- W3096593047 cites W2976228896 @default.
- W3096593047 cites W2981030070 @default.
- W3096593047 cites W3014488508 @default.
- W3096593047 cites W3032919895 @default.
- W3096593047 cites W3038194455 @default.
- W3096593047 cites W3101442004 @default.
- W3096593047 cites W3105609823 @default.
- W3096593047 hasPublicationYear "2020" @default.
- W3096593047 type Work @default.
- W3096593047 sameAs 3096593047 @default.
- W3096593047 citedByCount "3" @default.
- W3096593047 countsByYear W30965930472021 @default.
- W3096593047 crossrefType "posted-content" @default.
- W3096593047 hasAuthorship W3096593047A5015524477 @default.
- W3096593047 hasAuthorship W3096593047A5020531381 @default.
- W3096593047 hasAuthorship W3096593047A5036766747 @default.
- W3096593047 hasAuthorship W3096593047A5084848670 @default.
- W3096593047 hasConcept C107673813 @default.
- W3096593047 hasConcept C111919701 @default.
- W3096593047 hasConcept C11413529 @default.
- W3096593047 hasConcept C119857082 @default.
- W3096593047 hasConcept C121332964 @default.
- W3096593047 hasConcept C127705205 @default.
- W3096593047 hasConcept C134306372 @default.
- W3096593047 hasConcept C154945302 @default.
- W3096593047 hasConcept C17816587 @default.
- W3096593047 hasConcept C33923547 @default.
- W3096593047 hasConcept C36503486 @default.
- W3096593047 hasConcept C39920418 @default.
- W3096593047 hasConcept C41008148 @default.
- W3096593047 hasConcept C74650414 @default.
- W3096593047 hasConcept C90509273 @default.
- W3096593047 hasConcept C97541855 @default.
- W3096593047 hasConceptScore W3096593047C107673813 @default.
- W3096593047 hasConceptScore W3096593047C111919701 @default.
- W3096593047 hasConceptScore W3096593047C11413529 @default.
- W3096593047 hasConceptScore W3096593047C119857082 @default.
- W3096593047 hasConceptScore W3096593047C121332964 @default.
- W3096593047 hasConceptScore W3096593047C127705205 @default.
- W3096593047 hasConceptScore W3096593047C134306372 @default.
- W3096593047 hasConceptScore W3096593047C154945302 @default.
- W3096593047 hasConceptScore W3096593047C17816587 @default.
- W3096593047 hasConceptScore W3096593047C33923547 @default.
- W3096593047 hasConceptScore W3096593047C36503486 @default.
- W3096593047 hasConceptScore W3096593047C39920418 @default.
- W3096593047 hasConceptScore W3096593047C41008148 @default.
- W3096593047 hasConceptScore W3096593047C74650414 @default.
- W3096593047 hasConceptScore W3096593047C90509273 @default.
- W3096593047 hasConceptScore W3096593047C97541855 @default.
- W3096593047 hasLocation W30965930471 @default.
- W3096593047 hasOpenAccess W3096593047 @default.
- W3096593047 hasPrimaryLocation W30965930471 @default.
- W3096593047 hasRelatedWork W2128535275 @default.
- W3096593047 hasRelatedWork W2529477964 @default.
- W3096593047 hasRelatedWork W2790341021 @default.
- W3096593047 hasRelatedWork W2803429660 @default.
- W3096593047 hasRelatedWork W2886730169 @default.
- W3096593047 hasRelatedWork W2948443723 @default.
- W3096593047 hasRelatedWork W2951948137 @default.
- W3096593047 hasRelatedWork W2962812027 @default.
- W3096593047 hasRelatedWork W2964278684 @default.
- W3096593047 hasRelatedWork W3009795828 @default.
- W3096593047 hasRelatedWork W3010612407 @default.
- W3096593047 hasRelatedWork W3011973854 @default.
- W3096593047 hasRelatedWork W3088347926 @default.
- W3096593047 hasRelatedWork W3102701697 @default.
- W3096593047 hasRelatedWork W3104329975 @default.
- W3096593047 hasRelatedWork W3119351950 @default.
- W3096593047 hasRelatedWork W3126300335 @default.
- W3096593047 hasRelatedWork W3164133736 @default.
- W3096593047 hasRelatedWork W3206938627 @default.
- W3096593047 hasRelatedWork W3209810522 @default.
- W3096593047 isParatext "false" @default.
- W3096593047 isRetracted "false" @default.
- W3096593047 magId "3096593047" @default.
- W3096593047 workType "article" @default.