FlightBench
收藏资源简介:
FlightBench是由清华大学开发的首个针对四旋翼无人机的3D空间规划综合开源基准。该数据集旨在比较经典优化方法与新兴学习方法在复杂环境中的空间规划能力。数据集包含多种难度级别的任务,通过创新的任务难度和性能指标进行量化评估。FlightBench的应用领域主要集中在提升四旋翼无人机在高速飞行和实时规划中的性能,特别是在处理复杂条件如大角度转弯或视线遮挡时的能力。
FlightBench is the first comprehensive open-source benchmark dataset for 3D spatial planning of quadrotor unmanned aerial vehicles (UAVs), developed by Tsinghua University. This benchmark aims to compare the spatial planning capabilities of classical optimization-based methods and emerging learning-based methods in complex environments. It encompasses tasks across multiple difficulty levels, and conducts quantitative evaluations through innovative task difficulty metrics and performance indicators. The primary application domains of FlightBench focus on improving the performance of quadrotor UAVs in high-speed flight and real-time planning, particularly their ability to handle complex conditions such as sharp turns and line-of-sight occlusions.




