中文摘要：

随着智能规划的发展,其所面对的问题规模越来越大,而且可以预见以后会更大.现有的研究大多用二级存储扩展空间,其终极形式应该是用数据库进行存储.此外,有很多同一领域的规划问题,其所包含的常量几乎一致,其中必然有可重用信息来帮助加速求解.要更好地利用这些可重用信息也需要数据库.考虑到以上两个问题,首次提出规划领域描述语言PDDL（planning domain description language）的ER 模型（entity relationship model）,并基于此模型用存储过程来编写规划器SPP（stored procedure planner）.SPP 是完全在数据库内部运行的最优规划器,存取效率高,可充分利用数据库的各种功能.在国际规划大赛IPC（Int’l planning competition）基准领域上的实验结果表明,在有限的机器配置下,SPP 可以求解传统最优规划器不能求解的问题.该工作迈出了在数据库中求解规划问题,从而彻底解决空间问题的第一步.

英文摘要：

With the development of automated planning, the size of problems is getting bigger and bigger, and one can predict that it will become very large in the future. Some existing research work begins to use secondary memories to extend the search space, and it is believed databases are finally used. Besides, many problems that belong to the same domain often have common constants, so there might be a plenty of reusable information to speed up the solution process. To store this information permanently, databases are also needed. Inspired by the above two reasons, this paper first proposes ER （entity relationship） models for PDDL （planning domain description languages） and then develops a stored procedure based automated planner （stored procedure planner, SPP） for the first time. This planner is an optimal one which runs totally inside a database. It stores and accesses data efficiently and takes fully advantages of database features. Experiments on benchmark problems from the Int＇l planning competition （IPC） show that this planner can solve problems which cannot be solved by some classical optimal planners in a limited machine configuration. The work in this paper takes the first step for solving planning problems totally in a database so that it is helpful to solve huge-size problems finally.