Speaking the Navigator's Language: Trajectory-Grounded Instruction Translation for Frozen Aerial VLN Agents

مقتطف من المصدر

تنظم هذه الصفحة محتوى عن الذكاء الاصطناعي والبحث لتسهيل قراءته والاستشهاد به والتحقق منه من قبل الباحثين والطلاب وأدوات الذكاء الاصطناعي.

arXiv:2610.10635v1 Announce Type: new Abstract: Aerial vision-and-language navigation (VLN) agents are typically trained on detail-rich, trajectory-aligned commands, whereas users issue short, intent-driven instructions; on a frozen OpenFly navigator, this \emph{instruction gap} drops success rate (SR) from $31.03\%$ to $11.33\%$. To scale translator training, we prompt a language model with human-written style examples to convert original commands into paired, intent-centered Weak commands, which yield $15.27\%$ SR. We introduce the \textbf{Trajectory-Grounded Instruction Translator (TGIT)}, a front-end that keeps the navigator frozen and translates Weak inputs into agent-executable commands by learning from its trajectory outcomes. The resulting Weak-trained translator raises Weak-input SR to $37.93\%$ and transfers zero-shot to real human instructions ($11.33\%{\rightarrow}32.51\%$); it also improves held-out OpenFly ($4.95\%{\rightarrow}20.79\%$) and yields recovery on CityNav and AirVLN.

intelligence artificielle

عرض المصدر الأصلي

المصدر الرسمي أو الأصلي: arXiv AI. تحقق دائما من التفاصيل في المصدر الأساسي.

العودة إلى الذكاء الاصطناعي والبحث