Logical Regression for Planning with Axioms

2026-07-23

AI Summary

This work addresses the challenge of efficiently deriving the most general preconditions required to achieve a goal in automated planning with axioms, a task where traditional logical regression suffers from high computational complexity. To overcome this limitation, the paper proposes an approximate logical regression method that restricts preconditions to partial states, thereby avoiding redundant axiom computations while efficiently generating minimal partial states. This approach constitutes the first efficient approximation of logical regression in axiom-rich planning domains, significantly enhancing the generalization capability of partial states and the robustness of execution monitoring. Empirical results across multiple planning domains demonstrate up to a 70% reduction in the number of monitored variables and over 50% task recovery success under unexpected environmental perturbations.

Abstract

In automated planning, logical regression is an operation that returns the most general condition necessary for an action to achieve a particular formula. It has many applications, such as allowing for more robust plan execution and providing compact policies for non-deterministic planning. Although relatively simple to calculate in basic planning settings, logical regression becomes significantly more complex when additional factors, such as axioms, are present. 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 benefit greatly 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%, and demonstrate that 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.

Problem

• logical regression
• automated planning
• axioms
• execution monitoring
• partial states

Innovation

• logical regression
• axioms
• partial states
• execution monitoring
• automated planning