POMDP Benchmark Environments

该数据集包含了一系列标准的POMDP基准测试，包括老虎问题、岩石采样、阿罗哈、标签、老虎网格、走廊1、走廊2、五角大楼和第四等环境，同时还新增了如网格导航问题以及一个名为K-out-of-N的维护环境。此外，数据集还包含了POMDP求解器中使用的各种边界实现细节及比较。这些环境的规模从2到12545个状态，3到29个动作，以及1到1052个观察值不等。该数据集的任务是评估POMDP求解器的上值边界及其性能。

This dataset includes a set of standard POMDP benchmark environments, namely Tiger Problem, Rock Sample, Aloha, Tag, Tiger Grid, Corridor 1, Corridor 2, Pentagon, and Fourth. It also adds new environments such as the Grid Navigation Problem and a maintenance environment named K-out-of-N. In addition, the dataset contains detailed implementations and comparative analyses of various bounds utilized in POMDP solvers. The scales of these environments vary, with state counts ranging from 2 to 12545, action counts from 3 to 29, and observation counts from 1 to 1052. The core task of this dataset is to evaluate the upper value bounds and the overall performance of POMDP solvers.