SeFA-Policy: Fast and Accurate Visuomotor Policy Learning with Selective Flow Alignment

作者：Rong Xue, Jiageng Mao, Mingtong Zhang, Yue Wang · 单位：University of Southern California, Meta-World (easy), Meta-World (hard), Meta-World (very · 会议/期刊：ICRA 2026 · 日期：2025-11-11 · 来源：Low-Level Learning-Based Action Modelling / Policy Learning / Flow Matching Policy