FreeFly-Thinking : Aligning Chain-of-Thought Reasoning with Continuous UAV Navigation

Jiaxu Zhou; Shaobo Wang; Zhiyuan Yang; Zhenjun Yu; Tao Li

cs.CVMar 7, 2026

FreeFly-Thinking : Aligning Chain-of-Thought Reasoning with Continuous UAV Navigation

Authors: Jiaxu Zhou, Shaobo Wang, Zhiyuan Yang, Zhenjun Yu, Tao Li

Organizations: SJTU Paris Elite Institute of Technology, Shanghai Jiao Tong University, China · School of Physics and Astronomy, Shanghai Jiao Tong University, China · Tongji University, China

Abstract

Vision-Language Navigation aims to enable agents to understand natural language instructions and carry out appropriate navigation actions in real-world environments. Most work focuses on indoor settings, with little research in complex outdoor scenes. Current UAV Vision-and-Language Navigation models typically act as black boxes without explicit reasoning. We introduce FreeFly-thinking, an end-to-end VLN framework that converts the UAV agent's egocentric images and language instructions into a series of actions, inspired by environment of urban architecture proposed by OpenFly. We first construct a UAV dataset for navigation task, and then performing natural language chain of thought. We adopt a two-stage training strategy: Supervised fine-tuning and Reinforcement fine-tuning. Experiments on unseen test demonstrate a strong performance, presenting robustness and efficiency in UAV navigation issue.

Figures & tables

Figure 1: Overview for our work, illustrating the complete pipeline from automated Chain-of-Thought (CoT) data preparation based on visual observations and instructions, to the two-stage training paradigm (SFT and RFT) that optimizes the dual-head Freefly-thinking model for UAV navigation.

Figure 2: Architecture for Freefly model, which extracts shared multimodal representations from sequential visual observations and language instructions to simultaneously predict explicit Chain-of-Thought reasoning via a Language Model head and continuous 3D spatial waypoints via a Waypoint head

Panel of Figure 3 — Figure 3: Statistical overview of our constructed UAV navigation dataset. (a) The sample size distribution of primary flight maneuvers (e.g., straight, turn). (b) The distribution of physical trajectory lengths. (c) A word cloud visualization highlighting the most frequent terms in the generated Chain-of-Thought (CoT) rationales.

Method	NE $\downarrow$	SR $\uparrow$	ADE $\downarrow$
AerialVLN	45.9	4.3%	3.7
OpenFly	32.7	11.3%	2.6
Ours	28.0	13.1%	2.3

Table 1: Evaluation results on the unseen test set. NE and ADE are measured in meters (lower is better), while SR and Act. Acc are presented as percentages (higher is better).

Method	LM-head				Waypoint-head
	NE $\downarrow$	SR $\uparrow$	ADE $\downarrow$	Act. Acc $\uparrow$	NE $\downarrow$	SR $\uparrow$	ADE $\downarrow$
RFT	16.5	30.4%	1.4	84.5%	35.1	9.6%	3.2
SFT w/o thinking	19.1	27.2%	1.5	83.0%	31.8	11.0%	2.5
SFT dual head	18.7	27.6%	1.5	83.1%	28.0	13.1%	2.3

Table 2: Evaluation results for ablation experiments.

Aerial VLN

Aerial vision-and-language navigation (VLN) enables UAVs to follow natural-language goals using visual perception and flight control.

See all 61 papers

Think Like a Pilot: Fine-Grained Long-Horizon UAV Navigation
Jun 5, 2026Xiangyi Zheng, Xiangyu Wang, Qinan Liao +6UAV Navigation Vision-Language Navigation
Colab, Beihang University · National University of Singapore · Meituan
ImagineUAV: Aerial Vision-Language Navigation via World-Action Modeling and Kinodynamic Planning
May 31, 2026Xuchen Liu, Jiawei Huang, Shihao Xia +3UAV Navigation Vision-Language Navigation
Pengcheng Laboratory, Shenzhen, Guangdong, China · School of Computer Science and Cyber Engineering, Guangzhou University, Guangzhou, Guangdong, China · Southern University of Science and Technology, Shenzhen, Guangdong, China
FineCog-Nav: Integrating Fine-grained Cognitive Modules for Zero-shot Multimodal UAV Navigation
Apr 17, 2026Dian Shao, Zhengzheng Xu, Peiyang Wang +4Zero-Shot Learning UAV Navigation
Northwestern Polytechnical University · Nanjing University

Jun 5, 2026J/K move · Enter open · S save

Think Like a Pilot: Fine-Grained Long-Horizon UAV Navigation

Xiangyi Zheng, Xiangyu Wang, Qinan Liao, Zimu Tang+5

Colab, Beihang University · National University of Singapore · Meituan

UAV Navigation Vision-Language Navigation VLM Reasoning Vision-Language-Action Models Aerial VLN

Open paper AskPDF arXiv

Language-guided UAV agents must execute long-horizon semantic instructions while producing smooth, physically feasible continuous flight commands, yet existing Vision-Language Navigation (VLN) benchmarks typically use discrete or coarse actions and existing UAV Vision-Language-Action (VLA) tasks focus on short, atomic maneuvers. To address this gap in UAV task settings, we introduce FLIGHT, a Fine-grained Long-horizon Instruction-Guided benchmark for Hybrid UAV navigation and reasoning Tasks, which combines multi-stage instructions with dense 6-DoF trajectory annotations across two dataset splits: Fine-grained VLN and Long-horizon Flow. To endow the UAV agent with the capability of real-time in-flight reasoning over task execution status and mission planning, while simultaneously accommodating high-frequency, real-time precise control, we further propose FLIGHT VLA, an asynchronous architecture that decouples a low-frequency Streaming Pilot Vision-Language Model (VLM) for task-state reasoning from a high-frequency diffusion action model for continuous control, supervised by explicit Pilot Reasoning texts that summarize the current flight state and anticipate the next subgoal. In closed-loop evaluation, FLIGHT VLA consistently surpasses representative VLN and VLA baselines on our FLIGHT benchmarks, achieving stronger multi-stage completion, subgoal adherence, and terminal control. Its trained Streaming Pilot Reasoning VLM further improves UAV video reasoning, validating the effectiveness of our design.

Abstract

Aerial VLN

Explore similar work

Think Like a Pilot: Fine-Grained Long-Horizon UAV Navigation

ImagineUAV: Aerial Vision-Language Navigation via World-Action Modeling and Kinodynamic Planning

FineCog-Nav: Integrating Fine-grained Cognitive Modules for Zero-shot Multimodal UAV Navigation