Real-World Badminton Trajectories

该数据集包含了从现实实验中收集的1万个原始羽毛球轨迹数据，其中9000个通过数据增强用于训练，而1000个则作为测试数据集。该数据集还包括了评估机器人与羽毛球发球机器及人类球员对弈时的表现指标，如击球率和成功率。规模上，数据集包含了1万个原始轨迹，其任务是用以训练基于学习的羽毛球机器人控制策略网络。

This dataset contains 10,000 raw badminton trajectory samples collected from real-world experiments. Specifically, 9,000 of these samples are utilized for model training after data augmentation, while the remaining 1,000 samples are reserved as the test dataset. Additionally, the dataset encompasses performance metrics for evaluating robotic badminton systems competing against both automated badminton serving machines and human players, including shot hit rate and success rate. With a total scale of 10,000 raw trajectories, this dataset is designed for training learning-based badminton robot control policy networks.