Object Manipulation Environment

该数据集模拟了一个包含六个房间的建筑平面图离散环境，其中智能体需要根据自然语言指令来捡起并运送物体。由于任务描述不够明确，因此生成了子目标以指导智能体的学习过程。数据集规模包括3240个可能的子目标列表，其中1700个用于训练。具体任务为：捡起一个物体并将其运送到指定位置。

This dataset simulates a discrete environment based on a building floor plan with six rooms, where an AI Agent is required to pick up and transport objects in accordance with natural language instructions. Given the ambiguity of the task descriptions, sub-goals are generated to guide the learning process of the AI Agent. The dataset comprises 3240 possible sub-goal lists, with 1700 of them allocated for training purposes. The specific task is defined as: pick up an object and transport it to a designated location.