论文/arXiv 2610.22092·13 天前

用于人类感知安全 UAV 导航的机载轻量级 LLM 推理

针对UAV将人、车、墙同作占据几何的安全缺口，提出机载BLIP与轻量LLM管线，并在Gazebo导航与着陆中验证。

AI 整理

• 本体 SDK 与控制
• 感知与导航
• 仿真与训练

原文链接：Onboard Lightweight LLM Reasoning for Human-Aware Safe UAV Navigation（arXiv 2610.22092 · 发布日期：2026-10-15）

Onboard Lightweight LLM Reasoning for Human-Aware Safe UAV Navigation

针对UAV将人、车、墙同作占据几何的安全缺口，提出机载BLIP与轻量LLM管线，并在Gazebo导航与着陆中验证。

核心要点

• 传统感知—规划把行人、停放车辆和墙都化为占据几何，缺少类别相关安全间距。
• 共享BLIP编码器进行场景描述和文本条件分割，并融合深度信息定位3D障碍物。
• 轻量LLM按目标类别推断风险，结合RAG和语法约束生成规划器兼容JSON，并缓存结论。
• Gazebo中碰撞事件较几何规划器约降3倍，语义感知着陆成功率由34%升至96%；224像素输入限制远距小目标感知。

验证范围与局限：验证限于Gazebo仿真中的低空城市导航以及移动参与者间的语义感知着陆任务。

英文摘要 (Abstract)

Aerial robots operating near people must treat a pedestrian, a parked car, and a wall differently, yet conventional perception--planning stacks reduce all of them to occupied geometry---a \emph{human-aware safety gap} in which category-specific clearance is unavailable to the planner and temporary occlusions remove pedestrian-specific constraints from its map. We present a modular, fully onboard pipeline that closes this gap by turning visual observations into context-dependent safety constraints for real-time UAV navigation. A shared BLIP encoder drives scene captioning and text-conditioned segmentation, which is fused with depth to localize obstacles in 3D. Obstacle-specific risk is inferred by a lightweight LLM (Qwen-2.5-0.5B) stabilized with retrieval-augmented generation and grammar-constrained decoding that restricts its output to a planner-compatible JSON schema. The LLM reasons once per object category rather than per frame: its verdicts are cached per class, so previously seen categories are resolved without inference and new ones trigger a fresh retrieval-grounded query; detections are additionally given persistent identities and held in the map across occlusions. The perception pipeline runs at 10,Hz on a Jetson Orin NX, while fallback LLM queries take approximately 0.8,s and remain off the control-critical path. In Gazebo, low-altitude urban navigation among pedestrians and vehicles shows a roughly $3\times$ reduction in collision events over a geometric planner, and semantics-aware landing among moving actors raises success from 34% to 96%. The 224-pixel encoder input limits reliable small-object perception at range; conservative pedestrian margins and short-term persistence mitigate this limitation during navigation.

作者：Siwei Cai, Yuwei Wu, Vijay Kumar, Lifeng Zhou

来源

本页事实对应的公开页面，以原文为准。

1. arXiv 2610.22092：Onboard Lightweight LLM Reasoning for Human-Aware Safe UAV Navigation访问 2026-10-21