In automated planning, logical regression is an operation that returns the most general condition necessary for an action to achieve a particular formula. This technique demonstrates its importance across various applications, such as enhancing the robustness of plan execution and providing compact policies for non-deterministic planning.
While relatively simple to calculate in basic planning settings, logical regression becomes significantly more complex when additional factors, such as axioms, are introduced.
We introduce a methodology for approximating the logical regression of an action in a domain that includes axioms; an approximation that limits conditions to partial states. Our method produces minimal partial states while avoiding the recalculation of axioms.
To demonstrate the impact of our methods, we embed our form of regression in an execution monitoring context, a well-established setting that can greatly benefit from logical regression. Our results show that this form of regression can dramatically generalize partial states across multiple domains, reducing the number of variables considered for execution monitoring by up to 70%.
Furthermore, the resulting execution monitor is robust enough to recover frequently in an environment with unexpected changes: several domains recover over 50% of the time in our tests.
Blogger's Review: The logical regression method proposed in this paper showcases its powerful application potential in complex automated planning, significantly improving the efficiency and robustness of execution monitoring through effective handling of axioms.