Hallway POMDP Benchmark

该数据集是一个基准测试，其中机器人导航一个网格迷宫，感知墙壁的存在，并在某些单元格中尝试任务，这些任务有不同的结果概率。数据集还包括了不同风险阈值下的变体，以及多次执行后平均的结果。在规模上，该数据集的状态空间大小为67,584，动作数量为3，观察结果数量为43。该任务是在POMDP框架下进行的不确定性决策制定。

This dataset is a benchmark test in which a robot navigates a grid maze, detects the presence of walls, and attempts tasks in specific cells, where each task has varying outcome probabilities. The dataset also encompasses variants under different risk thresholds, alongside averaged results derived from multiple execution runs. In terms of scale, this dataset features a state space size of 67,584, with 3 possible actions and 43 distinct observations. This task focuses on uncertain decision-making within the POMDP framework.