TY - JOUR
T1 - Cross-Domain Policy Adaptation via Value-Guided Data Filtering
AU - Xu, Kang
AU - Bai, Chenjia
AU - Ma, Xiaoteng
AU - Wang, Dong
AU - Zhao, Bin
AU - Wang, Zhen
AU - Li, Xuelong
AU - Li, Wei
N1 - Publisher Copyright:
© 2023 Neural information processing systems foundation. All rights reserved.
PY - 2023
Y1 - 2023
N2 - Generalizing policies across different domains with dynamics mismatch poses a significant challenge in reinforcement learning. For example, a robot learns the policy in a simulator, but when it is deployed in the real world, the dynamics of the environment may be different. Given the source and target domain with dynamics mismatch, we consider the online dynamics adaptation problem, in which case the agent can access sufficient source domain data while online interactions with the target domain are limited. Existing research has attempted to solve the problem from the dynamics discrepancy perspective. In this work, we reveal the limitations of these methods and explore the problem from the value difference perspective via a novel insight on the value consistency across domains. Specifically, we present the Value-Guided Data Filtering (VGDF) algorithm, which selectively shares transitions from the source domain based on the proximity of paired value targets across the two domains. Empirical results on various environments with kinematic and morphology shifts demonstrate that our method achieves superior performance compared to prior approaches.
AB - Generalizing policies across different domains with dynamics mismatch poses a significant challenge in reinforcement learning. For example, a robot learns the policy in a simulator, but when it is deployed in the real world, the dynamics of the environment may be different. Given the source and target domain with dynamics mismatch, we consider the online dynamics adaptation problem, in which case the agent can access sufficient source domain data while online interactions with the target domain are limited. Existing research has attempted to solve the problem from the dynamics discrepancy perspective. In this work, we reveal the limitations of these methods and explore the problem from the value difference perspective via a novel insight on the value consistency across domains. Specifically, we present the Value-Guided Data Filtering (VGDF) algorithm, which selectively shares transitions from the source domain based on the proximity of paired value targets across the two domains. Empirical results on various environments with kinematic and morphology shifts demonstrate that our method achieves superior performance compared to prior approaches.
UR - http://www.scopus.com/inward/record.url?scp=85191187012&partnerID=8YFLogxK
M3 - 会议文章
AN - SCOPUS:85191187012
SN - 1049-5258
VL - 36
JO - Advances in Neural Information Processing Systems
JF - Advances in Neural Information Processing Systems
T2 - 37th Conference on Neural Information Processing Systems, NeurIPS 2023
Y2 - 10 December 2023 through 16 December 2023
ER -