Abstract
One of the main reasons why plan execution meets with failure is because the environment in which the plan is being executed does not conform to an agent's expectations of how it will behave. If an agent is forming and attempting to execute plans in a given domain, these plans will be based on the agent's understanding of the domain, described in its ontology. Errors in this ontology are likely to lead to plans that are not executable. We propose to address this problem by dynamically refining the ontology as execution failure occurs and replanning using this updated ontology, thus creating more robust plans that are more likely to be executable.
Original language | English |
---|---|
Publication status | Published - Jun 2003 |
Event | Proceedings of workshop on Plan Execution at ICAPS '03 - Trento, Italy Duration: 9 Jun 2003 → 13 Jun 2003 |
Conference
Conference | Proceedings of workshop on Plan Execution at ICAPS '03 |
---|---|
Country/Territory | Italy |
City | Trento |
Period | 9/06/03 → 13/06/03 |