Spaceship Dataset

该数据集是一类基于物理的任务集合，其中要求宇宙飞船在精确的11个时间步内与其母船会合，同时受到多个行星的引力影响。此外，该数据集被用于训练和评估不同类型的代理（反应型、迭代型和元控制器型）在解决宇宙飞船会合任务中的性能表现。数据集按规模分为五个版本，每个版本中宇宙飞船任务的行星数量各不相同。这些任务主要涉及基于物理的决策制定和优化。

This dataset consists of a collection of physics-based tasks, where spacecraft must rendezvous with their mothership within exactly 11 time steps while being subject to gravitational perturbations from multiple planets. Additionally, this dataset is employed to train and assess the performance of diverse agent types (reactive, iterative, and meta-controller agents) when solving the spacecraft rendezvous task. The dataset is divided into five scaled versions, each featuring a distinct number of planets for the spacecraft's mission. These tasks primarily center on physics-based decision-making and optimization.