WOFOSTGym

该数据集名为WOFOSTGym，是一个专为训练强化学习代理而设计的作物模拟环境。它旨在优化单一农场和多农场设置下年度作物和多年生作物的农业管理决策。该环境支持23种年度作物和两种多年生作物，使得在真实限制条件下，可以学习到多样化的农业管理策略。WOFOSTGym让研究人员能够在不具备农业专业知识的情况下探索农业管理问题，并适用于在真实农业限制条件下测试强化学习算法。此外，该数据集支持多种作物和土壤类型，其任务重点是作物管理决策制定。

This dataset, designated as WOFOSTGym, is a crop simulation environment purpose-built for training reinforcement learning agents. It is designed to optimize agricultural management decisions for both annual and perennial crops under single-farm and multi-farm scenarios. The environment supports 23 types of annual crops and two types of perennial crops, enabling the learning of diversified agricultural management strategies under realistic constraint conditions. WOFOSTGym enables researchers to investigate agricultural management issues without requiring specialized agricultural expertise, and is applicable for testing reinforcement learning algorithms under real-world agricultural constraints. Furthermore, this dataset supports a wide range of crop and soil types, with its tasks centered on crop management decision-making.