Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387394711> ?p ?o ?g. }
- W4387394711 endingPage "104550" @default.
- W4387394711 startingPage "104550" @default.
- W4387394711 abstract "Quadruped robots have attracted many researchers due to their unique advantages. In the field of control, deep reinforcement learning (DRL) saves the complex and tedious design of traditional control algorithms and motivates the robot to learn the motion patterns by itself, which is a promising alternative approach. However, due to the lack of prior information that is difficult to obtain in the physical world, policy usually exhibits a single motion state when faced with multiple scenarios. In this paper, we adopt a teacher-student policy architecture that uses the student policy to implicitly identify the teacher policy containing a privileged information, enabling the robot to adjust its motion depending on the environment it is in. We propose the synthetic terrain-command curriculum that assigns different levels of command curriculum depending on the difficulty of the terrain, which is useful for improving policy adaptation and stability. Policy is often trapped in local optima due to the nonlinearity of the robot system and unreasonable parameters. We propose a self-correction mechanism to jump out of this state in time, so that the policy can be steadily improved. We demonstrate our policy on the real quadruped platform named SDUQuad-48, and the results show that the learned policy exhibits state-of-the-art performance in both high-speed locomotion and complex terrain adaptation." @default.
- W4387394711 created "2023-10-07" @default.
- W4387394711 creator A5009366968 @default.
- W4387394711 creator A5018790712 @default.
- W4387394711 creator A5025820722 @default.
- W4387394711 creator A5029539712 @default.
- W4387394711 creator A5061179511 @default.
- W4387394711 creator A5069853141 @default.
- W4387394711 creator A5070648653 @default.
- W4387394711 creator A5071826905 @default.
- W4387394711 date "2023-12-01" @default.
- W4387394711 modified "2023-10-18" @default.
- W4387394711 title "Stable skill improvement of quadruped robot based on privileged information and curriculum guidance" @default.
- W4387394711 cites W1985161922 @default.
- W4387394711 cites W2010245514 @default.
- W4387394711 cites W2065238107 @default.
- W4387394711 cites W2296073425 @default.
- W4387394711 cites W2539534359 @default.
- W4387394711 cites W2773059370 @default.
- W4387394711 cites W2793991136 @default.
- W4387394711 cites W2909553221 @default.
- W4387394711 cites W2911087563 @default.
- W4387394711 cites W2968717280 @default.
- W4387394711 cites W2968764556 @default.
- W4387394711 cites W3001997350 @default.
- W4387394711 cites W3010768390 @default.
- W4387394711 cites W3013699912 @default.
- W4387394711 cites W3080662046 @default.
- W4387394711 cites W3090033167 @default.
- W4387394711 cites W3100789280 @default.
- W4387394711 cites W3101442004 @default.
- W4387394711 cites W3104876774 @default.
- W4387394711 cites W3113616534 @default.
- W4387394711 cites W3117372156 @default.
- W4387394711 cites W3120778962 @default.
- W4387394711 cites W3177184855 @default.
- W4387394711 cites W3199990163 @default.
- W4387394711 cites W3204535492 @default.
- W4387394711 cites W3207111613 @default.
- W4387394711 cites W4205430897 @default.
- W4387394711 cites W4206589784 @default.
- W4387394711 cites W4226143977 @default.
- W4387394711 cites W4241459016 @default.
- W4387394711 cites W4285203349 @default.
- W4387394711 doi "https://doi.org/10.1016/j.robot.2023.104550" @default.
- W4387394711 hasPublicationYear "2023" @default.
- W4387394711 type Work @default.
- W4387394711 citedByCount "0" @default.
- W4387394711 crossrefType "journal-article" @default.
- W4387394711 hasAuthorship W4387394711A5009366968 @default.
- W4387394711 hasAuthorship W4387394711A5018790712 @default.
- W4387394711 hasAuthorship W4387394711A5025820722 @default.
- W4387394711 hasAuthorship W4387394711A5029539712 @default.
- W4387394711 hasAuthorship W4387394711A5061179511 @default.
- W4387394711 hasAuthorship W4387394711A5069853141 @default.
- W4387394711 hasAuthorship W4387394711A5070648653 @default.
- W4387394711 hasAuthorship W4387394711A5071826905 @default.
- W4387394711 hasConcept C104114177 @default.
- W4387394711 hasConcept C107457646 @default.
- W4387394711 hasConcept C112972136 @default.
- W4387394711 hasConcept C11413529 @default.
- W4387394711 hasConcept C119857082 @default.
- W4387394711 hasConcept C120665830 @default.
- W4387394711 hasConcept C121332964 @default.
- W4387394711 hasConcept C123657996 @default.
- W4387394711 hasConcept C139807058 @default.
- W4387394711 hasConcept C142362112 @default.
- W4387394711 hasConcept C153349607 @default.
- W4387394711 hasConcept C154945302 @default.
- W4387394711 hasConcept C161840515 @default.
- W4387394711 hasConcept C17744445 @default.
- W4387394711 hasConcept C18903297 @default.
- W4387394711 hasConcept C199539241 @default.
- W4387394711 hasConcept C202444582 @default.
- W4387394711 hasConcept C2775924081 @default.
- W4387394711 hasConcept C33923547 @default.
- W4387394711 hasConcept C34413123 @default.
- W4387394711 hasConcept C41008148 @default.
- W4387394711 hasConcept C47177190 @default.
- W4387394711 hasConcept C48103436 @default.
- W4387394711 hasConcept C86803240 @default.
- W4387394711 hasConcept C90509273 @default.
- W4387394711 hasConcept C9652623 @default.
- W4387394711 hasConcept C97541855 @default.
- W4387394711 hasConceptScore W4387394711C104114177 @default.
- W4387394711 hasConceptScore W4387394711C107457646 @default.
- W4387394711 hasConceptScore W4387394711C112972136 @default.
- W4387394711 hasConceptScore W4387394711C11413529 @default.
- W4387394711 hasConceptScore W4387394711C119857082 @default.
- W4387394711 hasConceptScore W4387394711C120665830 @default.
- W4387394711 hasConceptScore W4387394711C121332964 @default.
- W4387394711 hasConceptScore W4387394711C123657996 @default.
- W4387394711 hasConceptScore W4387394711C139807058 @default.
- W4387394711 hasConceptScore W4387394711C142362112 @default.
- W4387394711 hasConceptScore W4387394711C153349607 @default.
- W4387394711 hasConceptScore W4387394711C154945302 @default.
- W4387394711 hasConceptScore W4387394711C161840515 @default.
- W4387394711 hasConceptScore W4387394711C17744445 @default.