cs.CVNov 18, 2026

AeroRSI: Recursive Self-Improvement for Aerial Vision-Language Navigation

Authors: Shiwei Wu, Yuwen Pan, Zhi Li, Yaowen Li, Yan Zhou, Xuemiao Xu, Yu Liu

Organizations: SIGS, Tsinghua University · Department of Electronic Engineering, Tsinghua University · Foshan University · South China University of Technology

Abstract

Vision-language-action (VLA) models enable unmanned aerial vehicles (UAVs) to follow natural-language instructions, yet still fail frequently in closed-loop flight. Correcting these failures requires repairs, i.e., actions that avoid them at the states the policy itself reaches, which human pilots cannot label at scale and rule-based planners cannot supply reliably. In this paper, we present AeroRSI, a recursive self-improvement framework in which an aerial navigation policy iteratively learns from its own failures. Our key insight is that a failing policy should judge its own repairs by flying on from them. Judged in this way, its sampled actions and even rule-based proposals that hurt it when imitated directly become reliable supervision. Moreover, an episode terminates at its first failure, concealing subsequent failures until the earlier ones are repaired. AeroRSI thus renews its supervision in every round, anchoring the failures of the current policy, verifying repairs by its continuation, and distilling them while keeping its behavior elsewhere. On TravelUAV, three rounds raise the success rate of a supervised 2B-parameter policy from 46.4% to 67.5% and reduce collisions by half, surpassing the previous state of the art by 10.6 points from the same initialization without extra human labels or inference cost. Code and models are available at https://github.com/starsholic/AeroRSI.

Figures & tables

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. RefineFly: Failure-Aware Post-Training for Aerial Vision-Language Navigation

    Aug 10, 2026Boxiong Wang, Hui Kang, Geng Sun +3UAV NavigationVision-Language Navigation

  2. CLOSER-VLN: Closed-Loop Self-Verified Retrieval-Augmented Reasoning for Aerial Vision-Language Navigation

    Jun 24, 2026Shaoxuan Li, Xiangyu Dong, Xiaoguang Ma +3UAV NavigationVision-Language Navigation

  3. Towards Precise Intent-Aligned VLA Aerial Navigation via Expert-Guided GRPO

    Jun 1, 2026Tianyang Chen, Wenjun Li, Xin zhou +2UAV NavigationGroup Relative Policy Optimization