An Office Scenario Dataset for Benchmarking Observation-equivalent Entities

This dataset consists of simulated plans and observations of agents in an office environment. The unique property of this data set compared to other datasets is that multiple agents are acting in the same space, which cannot be distinguished given the observations (i.e. the observations are non-identifying). For each of 36 different initial settings, 20 plan have been randomly sampled, i.e. this dataset consists of 720 plans in total. Additionally there is a webservice, that enables you to generate your own dataset.