Safety and recovery improve planning for household AI agents
SAGE: Symbolic Action-Gating and Editing for LLM Task Planners
RoboticsArtificial Intelligence
Summary
Household robots that use language models often plan actions without checking if those actions really make sense in their environment, which can cause mistakes. The authors present SAGE, a method that adds a simple safety check to block impossible actions and edits plans locally when something goes wrong. This makes planning safer and more efficient, especially in complex tasks or when unexpected errors happen during execution. Their tests show SAGE helps robots recover from failures more reliably while using fewer calls to the language model.
What this means in practice
- •For robotics software developers: Integrate a low-overhead safety layer to block unsafe robot actions and enable efficient plan repair during task execution in home environments.
- •For automated system testers: Use symbolic gating to verify AI-generated plans before execution and reduce costly replanning in complex simulated scenarios.
Authors
Trung Minh Bui, JongSul Moon, YoungOuk Kim, Quang-Ngoc Phung, Se-Woong Jun, Dongin Shin
Abstract
Large language models (LLMs) are now the default cognitive core of embodied household agents, yet the plans they emit are rarely checked against a grounded model of the environment before execution, and the task-success they report is often measured on benchmarks so saturated that no method can be separated from another. We present SAGE (Symbolic Action-Gating and Editing), a single-LLM planner built from two lightweight mechanisms: a domain-agnostic symbolic gate (~250 lines of Python, zero tokens, $O(|π|)$) that blocks precondition-violating actions with typed reasons as a runtime safety monitor, and a local edit that regenerates only the failed sub-goal's suffix, keeping completed and untouched work intact; a hybrid seed+live memory store supports cold-start coverage. We evaluate under a leak-free protocol (leave-one-out retrieval) over five open-weight models and a 75-task AI2-THOR benchmark. On the standard benchmark goal-completeness saturates (52% of instances trivially solved) and SAGE ties strong hierarchical baselines. On a harder, method-agnostic multi-goal composition, SAGE's completeness lead re-emerges large (+0.06 to +0.23 across four models). Under injected mid-execution failures, SAGE recovers as reliably as whole-plan replanners at 2.4-3.3x fewer LLM calls. As a verify-before-execute gate, the symbolic monitor blocks unsafe actions before actuation and raises simulator-reported step-success for every planner tested (up to +0.11), a signal the verifier never sees (non-circular). Because the gate calls no model (0.008 ms/plan), it is a safety layer that runs essentially free on the edge: SAGE planning reproduces its quality on a Jetson AGX Orin, where small-model verification helps most. We release the benchmark, the leak-free protocol, the recovery and safety-gate harnesses, and a verifier-portability study (auto-induced on ALFWorld, 0.89 held-out).