SFB 527 Integration symbolischer und subsymbolischer Informationsverarbeitung in adaptiven sensomotorischen Systemen: Subproject D1, Knowledge representation for autonomous systems

Project Description

For overview of the purpose and goals of the whole project, see the SFB 527 home page.

Symbolic knowledge representation is a key component in intelligent autonomous sensomotoric systems. The goal of the subproject is to develop a unified framework for knowledge representation that support both the symbolic and subsymbolic components of the SFB 527 architecture and the cooperation between these components. Important forms of knowledge that many of the components in the system share are the spatial description of the environment (shared e.g. by path planning components of different abstraction levels, as well as components for self-localization that are based on sensor input), the description of the sensomotoric system itself (interpretation of sensor input, lower levels of path and motion planning), and knowledge related to the tasks the sensomotoric system is carrying out. The requirement for fast response many of the afore-mentioned subtasks pose is an important topic for the research in the subproject.

Group Members

Publications

J. Rintanen. Complexity of prioritized default logics. Journal of Artificial Intelligence Research 9:423-461, 1998.

J. Rintanen and H. Jungholt. Numeric state variables in constraint-based planning, Recent Advances in AI Planning: 5th European Conference on Planning (ECP'99), S. Biundo, ed., Durham, England, Lecture Notes in Artificial Intelligence, September 1999. Springer-Verlag, Berlin, Germany.

J. Rintanen. Improvements to the evaluation of quantified Boolean formulae, Proceedings of the 16th International Joint Conference on Artificial Intelligence, T. Dean, ed., pages 1192-1197, Stockholm, Sweden, August 1999. Morgan Kaufmann Publishers, San Francisco, California.