arXiv cs.AI· Ruibin Ju, Ruifeng Cai, Min Xu, Lin Wang·· 2 天前AI 评分26
X-Fly：全块式空中视觉语言导航中的到达与停止学习
X-Fly: Learning to Reach and Stop in Full-Chunk Aerial Vision-Language Navigation
AI 导读
X-Fly 是一个面向全块式空中视觉语言导航的分阶段训练框架，通过 Recovery DAgger 为学习者访问的决策状态标注可重连未来专家片段的平滑动作块，并在首个区域内状态给出 STOP，再用 Flow-GRPO 联合优化分块流控制器与 Bernoulli 停止策略。
正文 · 原文
Abstract:Aerial vision-language navigation (VLN) agents built on large vision-language models are expensive to query. Executing a complete predicted action chunk before replanning amortizes this cost, reducing model inference from 195 to 42 ms per executed waypoint for our policy. It also moves the states at which the policy is queried away from expert demonstrations, which supervise neither recovery from these states nor termination at the learner's own arrival. On OpenFly test-seen, no-stop failures account for 47.78% of full-chunk behavior cloning (BC) outcomes. We present X-Fly, a staged training framework for full-chunk aerial VLN. Recovery DAgger labels learner-visited decision states with smooth chunks that reconnect to future expert segments and with STOP at the first in-region state. Flow-GRPO then jointly refines the chunked flow controller and a Bernoulli STOP policy from same-task closed-loop outcomes, supplemented by outcome-verified stopping supervision. With eight controls per decision, X-Fly reaches 74.18% SR on test-seen and 51.73% on test-unseen, compared with 16.45% and 18.43% for full-chunk BC and 32.04% and 29.46% for DreamFly.
| Comments: | 13 pages, 5 figures, 3 tables. Project page: this https URL |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.21450 [cs.RO] |
| (or arXiv:2610.21450v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.21450 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ruibin Ju [view email]
[v1]
Thu, 15 Oct 2026 07:54:18 UTC (1,428 KB)
来源：arXiv cs.AI · arxiv.org