SafeInferCom: Safe Inference-Time Compute via Verifier-Guided Mid-Generation Intervention for Robotic Task Planning

Weizhe Xu; Jialiang Fan; Mengyu Liu; Fanxin Kong

cs.ROOct 8, 2026

SafeInferCom: Safe Inference-Time Compute via Verifier-Guided Mid-Generation Intervention for Robotic Task Planning

Authors: Weizhe Xu, Jialiang Fan, Mengyu Liu, Fanxin Kong

Organizations: University of Notre Dame · Washington State University

Abstract

Large Reasoning Language Models (LRLMs) enable multi-step reasoning for robotic task planning, but continued reasoning can overwrite valid intermediate plans or leave constraint violations unresolved, reducing planning reliability and wasting inference-time computation. We develop an inference-time monitor that exposes and verifies intermediate plans without disrupting the original decoding trajectory. Building on this monitor, we propose SafeInferCom, a formal verifier-guided framework that preserves valid intermediate plans and directs error correction during generation. Experiments across multiple LRLMs and planning domains reveal reasoning-response inconsistency and limited self-correction under one-shot inference. SafeInferCom improves planning success and accelerates error correction relative to one-shot inference. When combined with iterative refinement, it further improves success while reducing token usage compared with refinement alone. We additionally evaluate SafeInferCom in VirtualHome and provide a real-world robotic-arm demonstration.

Figures & tables

Fig. 1: Overall workflow of the inference-time monitor. At a selected checkpoint, primary decoding is paused and an auxiliary branch appends the induction prompt $p_{\mathrm{ind}}$ to the reasoning prefix $\mathbf{r}_{<k}$ to produce an induced candidate plan $\pi_{k}$ . The external verifier returns its label $y_{k}$ , after which the auxiliary branch is discarded and primary decoding resumes until the final plan $\pi$ is produced.

Fig. 2: The workflow of our method SafeInferCom.

Method	Int. Plan	Verifier	Intervention	Early Stop
One	✗	✗	✗	✗
IR	✗	✓	Post-hoc	✗
Early	✓	✓	✗	✓
S	✓	✓	Mid-gen	✓
S+IR	✓	✓	Both	✓
FD	N/A	N/A	Search	N/A

TABLE I: Comparison of inference strategies. Int. denotes intermediate plan extraction. Verifier denotes the use of external verification within the solution procedure.

Metric	Model	BW	Log	Dep	Grip	Mic	Avg.
Suc. (%) $\uparrow$	7B	0.0	10.0	0.0	20.0	22.0	10.4
	14B	46.0	46.0	65.0	58.0	59.0	54.8
	32B	79.0	84.0	98.0	85.0	68.0	82.8
Tok. $\downarrow$	7B	5006	2277	2302	2479	1536	2720
	14B	4380	1730	2550	2797	2581	2808
	32B	2485	1464	1186	2768	1872	1955

TABLE II: Reasoning evaluation across LRLMs and planning domains

	7B		14B		32B
Method	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$
One	10.4	2720	54.8	2808	82.8	1955
IR	13.6	4861	83.2	4014	92.6	2290
Early	13.0	3075	60.0	2943	88.6	1920
S	14.2	3953	78.2	3220	93.2	2337
S+IR	19.6	4463	90.8	3952	98.4	2119

TABLE III: Final success rate (%) and token consumption (average).

	One	IR	S	S+IR
Mean	132.03	150.69	130.01	116.10
Median	109.69	119.58	95.59	83.61
P95	275.57	329.03	356.62	288.33

TABLE IV: End-to-end wall-clock time (s) for the 32B model.

Method	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$
	R1-14B		Qwen		Phi
One	46.0	1730	64.0	5186	6.0	6411
S	80.0	2460	67.0	4610	7.0	4943
S+IR	96.0	2778	97.0	7674	19.0	15587

TABLE V: Performance on additional LRLM families.

Method	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$	Succ. $\uparrow$	Tok. $\downarrow$
	7B		14B		32B
One	3.0	1489	20.0	1227	34.0	1282
S	8.0	7779	59.0	4411	52.0	3520
S+IR	16.0	20746	69.0	7823	61.0	7988

TABLE VI: Performance of SafeInferCom in VirtualHome.

Explore similar work

Using large language models for embodied planning introduces systematic safety risks
Apr 20, 2026Tao Zhang, Kaixian Qu, Zhibin Li +4Large Language Model-Based Robot Planning Robot Safety
ETH Zurich, Zurich, Switzerland. · University College London, London, United Kingdom. · Stanford University, Stanford, California, United States. +2
Agentic Harnesses: LLM-Driven Verification Layers for Robot Autonomy
Aug 10, 2026Rohan Bhagra, Mahantesh Halapannavar, Uddhav BhattaraiLarge Language Model-Based Robot Planning Adversarial Robustness
Carnegie Mellon University, Department of Information Systems
SafeRun: Enabling Determinism in LLM Planning for Running
Jun 8, 2026Meilin Chen, Zepeng Zhai, Jiaxuan Zhao +1LLM Planning LLM Safety Evaluation
Xiaohongshu Inc. · Institute of Information Engineering, Chinese Academy of Sciences

Apr 20, 2026, cs.AIJ/K move · Enter open · S save

Using large language models for embodied planning introduces systematic safety risks

Tao Zhang, Kaixian Qu, Zhibin Li, Jiajun Wu+3

ETH Zurich, Zurich, Switzerland. · University College London, London, United Kingdom. · Stanford University, Stanford, California, United States.+2

Large Language Model-Based Robot Planning Robot Safety Robot Task Planning LLM Safety Evaluation

Open paper AskPDF arXiv

Large language models are increasingly used as planners for robotic systems, yet how safely they plan remains an open question. To evaluate safe planning systematically, we introduce DESPITE, a benchmark of 12,279 tasks spanning physical and normative dangers with fully deterministic validation. Across 23 models, even near-perfect planning ability does not ensure safety: the best-planning model fails to produce a valid plan on only 0.4% of tasks but produces dangerous plans on 28.3%. Among 18 open-source models from 3B to 671B parameters, planning ability improves substantially with scale (0.4-99.3%) while safety awareness remains relatively flat (38-57%). We identify a multiplicative relationship between these two capacities, showing that larger models complete more tasks safely primarily through improved planning, not through better danger avoidance. Three proprietary reasoning models reach notably higher safety awareness (71-81%), while non-reasoning proprietary models and open-source reasoning models remain below 57%. As planning ability approaches saturation for frontier models, improving safety awareness becomes a central challenge for deploying language-model planners in robotic systems.

Abstract

Explore similar work

Using large language models for embodied planning introduces systematic safety risks

Agentic Harnesses: LLM-Driven Verification Layers for Robot Autonomy

SafeRun: Enabling Determinism in LLM Planning for Running