International Planning Competition (IPC) 2014 Domains
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/AI-Planning/classical-domains
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了一系列规划领域，如CityCar、Nurikabe、Maintenance、Briefcase、Miconic和Satellite，用于评估Conditional-SAM算法。数据集还包括了各种用于分类的特性，例如析取前提、存在前提、对象相等性、类型层次和蕴含关系。这些数据是使用PDDL问题生成器生成的，并通过五折交叉验证方法进行评估。每个领域生成了100个问题，其中Nurikabe领域解决了52个问题。该任务旨在从规划领域的轨迹中学习安全动作模型。
This dataset comprises a range of planning domains including CityCar, Nurikabe, Maintenance, Briefcase, Miconic, and Satellite, which are employed to evaluate the Conditional-SAM algorithm. It also incorporates various classification-related features such as disjunctive preconditions, existential preconditions, object equality, type hierarchies, and entailment relations. All data were generated using a PDDL problem generator and evaluated via five-fold cross-validation. One hundred problems were generated for each domain, among which 52 were successfully solved in the Nurikabe domain. The task aims to learn safe action models from trajectories within planning domains.
提供机构：
AI Planning Community
