Problem
Research questions and friction points this paper is trying to address.
planning task shielding
flaw detection
unsolvable planning tasks
goal specification
task repair
Planning Task Shielding: Detecting and Repairing Flaws in Planning Tasks through Turning them Unsolvable
🤖 AI Summary
This work proposes a novel paradigm termed “planning task shielding” to proactively prevent systems from entering hazardous states due to flaws in planning tasks. The approach transforms the original task into an unsolvable problem through minimal action intervention, thereby blocking defective execution paths. The core contribution is the design of allmin, an optimal algorithm grounded in classical planning formalism, which leverages off-the-shelf planners to detect task solvability and iteratively guides task reformulation with minimal modifications to achieve effective safety guarantees. Experimental results demonstrate that allmin operates efficiently and reliably across planning tasks of varying scales, successfully suppressing undesirable behaviors while strictly minimizing intervention, thus significantly enhancing system safety.
📝 Abstract
Most research in planning focuses on generating a plan to achieve a desired set of goals. However, a goal specification can also be used to encode a property that should never hold, allowing a planner to identify a trace that would reach a flawed state. In such cases, the objective may shift to modifying the planning task to ensure that the flawed state is never reached-in other words, to make the planning task unsolvable. In this paper we introduce planning task shielding: the problem of detecting and repairing flaws in planning tasks. We propose $allmin$, an optimal algorithm that solves these tasks by minimally modifying the original actions to render the planning task unsolvable. We empirically evaluate the performance of $allmin$ in shielding planning tasks of increasing size, showing how it can effectively shield the system by turning the planning task unsolvable.
Innovation
Methods, ideas, or system contributions that make the work stand out.
planning task shielding
unsolvable planning
flaw repair
minimal action modification
allmin algorithm
🔎 Similar Papers
No similar papers found.
