Adaptive Constraint-Based Agents in Artificial Environments

[INCOMPLETE KNOWLEDGE]   [Single Plan]   [Missing Information]   [Information Gathering]   [Partial Knowledge]

[ Please note: The project has been discontinued as of May 31, 2005 and is superseded by the projects of the ii Labs. There won't be further updates to these pages. ]

Missing Information

(Related publications: [PUBLink] [PUBLink])

The state resources can represent the lack of information by the addition of an Unknown value to their domain (which gets the default state). For example, in the agent's absence, other agents might open or close a door without the agent's noticing. Whether the door is closed or not can only be known if the door is within the agent's field of vision. Thus, the state resource Door with a domain of Open and Closed gets an additional Unknown value, which is triggered by actions that cause the agent's absence.

An action of passing the door requires that the Door is open. In a pessimistic approach, the precondition task of the passing action entails a bad satisfaction of the state resource if the Door is in an Unknown state because failure could endanger later commitments. The inclusion of a prophylactic Open Door action would avert this threat (see figure below).

On the other hand, the satisfaction could be driven by experience. If the agent has learned that the Door is usually open, the state resource Door might settle for the Unknown state with respect to a precondition check if the door is open.

[INCOMPLETE KNOWLEDGE]   [Single Plan]   [Missing Information]   [Information Gathering]   [Partial Knowledge]

For questions, comments or suggestions, please contact us.

Last update:
May 20, 2001 by Alexander Nareyek