Sparse Meta-world Tasks

该数据集包含了来自Meta-world MT10环境的十个机器人任务，这些任务以状态为基础进行观察，并采用稀疏奖励机制。当智能体达到目标时，会获得1的奖励，否则奖励为0。此外，该数据集被用于评估FuRL方法在强化学习任务中相较于其他各种基准方法的有效性。该数据集的任务规模为10个，主要针对的是强化学习任务。

This dataset contains ten robotic tasks sourced from the Meta-world MT10 environment, which adopt state-based observations and a sparse reward mechanism. Specifically, the agent will receive a reward of 1 when it achieves the target, and 0 otherwise. Furthermore, this dataset is utilized to evaluate the effectiveness of the FuRL method compared to various baseline methods in reinforcement learning tasks. This dataset consists of 10 tasks and is mainly targeted at reinforcement learning tasks.