MuJoCo Simulated Tasks

该数据集由四种类型的任务组成，这些任务通过MuJoCo模拟器进行评估，涉及不同类型的代理尝试到达不同的目标位置。具体包括：一个轮式代理、一个四足蚂蚁、一个双连杆伸长器和一个四连杆伸长器。此外，针对每个任务，基于目标分布设定了两种测试场景。在训练和测试过程中，每个任务都采样了100个目标位置。这些任务都属于强化学习任务，专注于不同代理到达目标位置的场景。

This dataset comprises four types of tasks evaluated using the MuJoCo physics simulator, where distinct agents attempt to navigate to various target positions. The specific agents include a wheeled agent, a quadrupedal ant, a two-link extender, and a four-link extender. Additionally, two test scenarios are established for each task based on the target distribution. During both training and testing phases, 100 target positions are sampled for every individual task. All these tasks belong to reinforcement learning tasks, focusing on the scenario where various agents reach their designated target positions.