OpenUAV
收藏资源简介:
OpenUAV数据集是由北京航空航天大学人工智能研究所创建的,专门用于无人机视觉语言导航任务。该数据集包含约12,000条轨迹,涵盖了6自由度的飞行动态,准确捕捉了无人机的复杂飞行行为。数据集的创建过程包括在OpenUAV平台上进行连续飞行,并使用GPT-4生成目标描述,随后进行人工质量检查。该数据集主要应用于无人机在复杂环境中的导航任务,旨在提高无人机在现实世界中的导航精度和效率。
The OpenUAV dataset was developed by the Institute of Artificial Intelligence at Beihang University, specifically tailored for unmanned aerial vehicle (UAV) vision-language navigation tasks. This dataset contains approximately 12,000 trajectories, covering 6-degree-of-freedom flight dynamics, and accurately captures the complex flight behaviors of UAVs. The development process of the dataset involves conducting continuous flight experiments on the OpenUAV platform, generating target descriptions with GPT-4, followed by manual quality inspection. It is mainly applied to UAV navigation tasks in complex environments, with the goal of enhancing the navigation accuracy and efficiency of UAVs in real-world scenarios.

- 1Towards Realistic UAV Vision-Language Navigation: Platform, Benchmark, and Methodology北京航空航天大学人工智能研究所 · 2024年



