Quadcopter Controller Training Data

该数据集包含了四旋翼无人机控制器在多种正常和非正常条件下的训练和评估结果，其中包括了部分电机故障和风速突增的情况。每个实验包含30个检查点，这些检查点的性能指标被汇总以供分析，每个检查点大约产生60MB的追踪数据。该数据集的规模宏大，240个经过训练的模型各自针对100个查询进行了评估，产生了大量的性能指标和追踪数据。这项任务是通过强化学习来实现四旋翼无人机的姿态控制。

This dataset encompasses training and evaluation results for quadrotor UAV controllers under diverse normal and abnormal operating conditions, including scenarios involving partial motor failures and sudden wind gusts. Each experiment includes 30 checkpoints, whose performance metrics are aggregated for analytical purposes, with each checkpoint generating approximately 60 MB of tracking data. This dataset boasts a considerable scale: 240 trained models are each evaluated across 100 queries, yielding substantial volumes of performance metrics and tracking data. The task focuses on implementing attitude control for quadrotor UAVs via reinforcement learning.