Technical note 418
SRI INTERNATIONAL MENLO PARK CA
Pagination or Media Count:
The ability to act appropriately in dynamic environments is critical to the survival of all living creatures. For lower life forms, it seems that sufficient capability is provided by stimulus-response and feedback mechanisms. Higher life forms, however, must be able to anticipate the future and form plans of action to achieve their goals. Reasoning about action and plans can thus be seen as fundamental to the development of intelligent machines that are capable of dealing effectively with real-world problems. Researchers in artificial intelligence AI have long been concerned with this area of investigation 73. But, as with most of AI, it is often difficult to relate the different streams of research and to understand how one technique compares with others. Much of this difficulty derives from the varied and sometimes confused terminology and the great diversity of problems that arise in real-world planning. Indeed, there are few practical planning systems for which the class of appropriate applications can be clearly delineated. This article attempts to clarify some of the issues that are important in reasoning about actions and plans. As the field is still young, it would be premature to expect us to have a stable foundation on which to build a discipline of planning. Nevertheless, I hope that the following discussion contributes toward that objective and that it will help the reader to evaluate the pertinent literature.
- Structural Engineering and Building Technology