DiVeR: Decision-Critical Verifier Learning for VLA Test-Time Scaling

Seongheon Park; Heecheol Kim; Shulin Tian; Lilika Makabe; Namiko Saito; Katsushi Ikeuchi; Sharon Li; Yasuyuki Matsushita

cs.ROOct 4, 2026

DiVeR: Decision-Critical Verifier Learning for VLA Test-Time Scaling

Authors: Seongheon Park, Heecheol Kim, Shulin Tian, Lilika Makabe, Namiko Saito, Katsushi Ikeuchi, Sharon Li, Yasuyuki Matsushita

Organizations: University of Wisconsin–Madison · Microsoft Research Asia–Tokyo · Nanyang Technological University · The University of Osaka · The University of Tokyo

Abstract

Scaling robot data and model capacity has improved Vision-Language-Action (VLA) policies, but further progress is constrained by the high cost of robotic data. Verifier-guided test-time scaling offers an efficient alternative by sampling multiple action candidates and selecting the one most likely to lead to task success at inference time. Existing classification-based verifiers learn from trajectory-level outcomes but treat all visited states equally, even though their value for candidate discrimination can vary across a trajectory. At many states, plausible actions are similar and provide limited discrimination signal, while only a sparse set of decision-critical states admits meaningfully different actions that can substantially affect downstream outcomes. To address this, we propose DiVeR, which estimates decision criticality from the dispersion of sampled action representations. DiVeR then uses this signal to reweight verifier learning toward states where action selection is most consequential, without requiring step-level annotations or additional environment interaction. Across LIBERO, RoboCasa, and real-world experiments on a Franka Research 3 robot, DiVeR consistently improves task success through more effective verifier-guided action selection, while adding negligible verifier inference overhead.

Figures & tables

Figure 1: Not all states matter equally for verifier learning. (a) At routine states , sampled action candidates remain similar, so distinguishing among them is less important for downstream behavior. At decision-critical states , candidates diverge more strongly, making accurate selection more consequential for task outcomes. (b) Across 500 LIBERO-Long trajectories from a frozen $\pi_{0}$ policy, states with high decision criticality , measured by $u_{t}$ ( Equation 8 ), are sparse and temporally localized. These observations motivate focusing verifier learning on sparse decision-critical states where action selection matters most.

Figure 2: Overview of DiVeR . (a) At each visited state, we sample $K$ action candidates from the frozen VLA policy and estimate decision criticality $u_{t}$ from the dispersion of their action-expert representations. The scores are standardized within each trajectory and converted into decision weights $w_{t}$ , which are paired with the executed action representation $z_{t}$ and trajectory outcome $y_{t}$ to construct the training dataset $\mathcal{D}$ . (b) The verifier is trained with decision-weighted binary cross-entropy, assigning greater importance to states with larger $w_{t}$ . (c) At inference, the verifier scores $N$ sampled action candidates at each decision step and executes the highest-scoring action chunk.

Figure 3: Sampled action representations. Candidate representations are concentrated at a routine state (e.g., free-space motion) but more dispersed at a decision-critical state (e.g., before grasping).

Method	Policy: $\bm{\pi_{0}}$				Policy: $\bm{\pi_{0.5}}$
Method	$N=4$	$N=8$	$N=16$	$N=32$	$N=4$	$N=8$	$N=16$	$N=32$
Baseline	$83.5$				$93.5$
KDPE ( Rosasco et al., 2025 )	$83.5$	$84.5$	$82.8$	$82.3$	$94.0$	$92.7$	$94.5$	$95.7$
MG-Select ( Jang et al., 2026 )	$82.0$	$80.8$	$83.3$	$86.3$	$94.2$	$95.0$	$\underline{96.5}$	$96.8$
TACO ( Yang et al., 2025 )	$\mathbf{84.2}$	$83.0$	$83.5$	$85.0$	$\underline{94.5}$	$94.8$	$95.5$	$95.7$
SVM ( Tsao et al., 2026 )	$82.8$	$\underline{85.2}$	$\underline{85.5}$	$\underline{86.5}$	$94.0$	$\underline{95.5}$	$94.0$	$\underline{97.0}$

Table 1: Average success rates (%) on LIBERO-Long with varying numbers of action candidates

N

sampled at test time. Bold and underlined values denote the best and second-best results, respectively.

Method	Policy: $\bm{\pi_{0}}$					Policy: $\bm{\pi_{0.5}}$
Method	Articulated	Control	Object-Centric	Navigation	Average	Articulated	Control	Object-Centric	Navigation	Average
Baseline	37.2	16.2	42.5	0.0	32.2	50.0	18.8	58.3	0.0	43.1
KDPE	32.8	26.2	49.2	0.0	35.0	47.8	20.0	57.5	0.0	43.3
MG-Select	29.3	22.5	36.7	0.0	28.6	53.6	33.8	66.7	15.0	51.3
TACO	37.1	28.8	35.8	0.0	32.7	52.1	26.2	53.3	0.0	43.9
SVM	37.2	26.2	40.8	0.0	33.9	55.0	20.0	60.8	15.0	46.9

Table 2: Category-wise average success rates (%) on RoboCasa atomic tasks. Bold and underlined values denote the best and second-best results, respectively.

Panel of Table 3 — Table 3: Real-robot evaluation with $\pi_{0.5}$ . Left : robot setup. Center : the four evaluation tasks. Right : task success rates (%) over 24 evaluation episodes per task.

Table 7

Figure 4: Cross-task generalization. Success rates (%) when the verifier is trained on one task category and evaluated on another.

Figure 5: Qualitative comparison of verifier-guided action selection. Starting from the same decision-critical state, we compare trajectories produced by a uniformly trained verifier and DiVeR (Ours) . In both (a) drawer closing and (b) block stacking, the two verifiers select different action candidates from the same state, leading to different downstream outcomes. Uniform verifier training selects actions that result in task failure, whereas DiVeR selects more appropriate candidates and successfully completes the task. Red circles highlight the resulting behaviors after these action selections.

Figure 6: Verifier architecture.

Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Hyperparameter	Search Space
Optimization
Optimizer	{Adam, AdamW , SGD}
Learning rate	{1e-2, 5e-3, 1e-3, 5e-4 , 1e-4}
Training epochs	{10, 30 , 50, 100}
Architecture
MLP hidden size	{ 64 , 256, 512}

Appendix

Table 7: Hyperparameter search space applied uniformly across all experiments. Selected values are bolded .

Table 8: Dataset statistics. Number of successful and failed episodes and their corresponding environment steps in the offline rollout datasets collected with (a)

\pi_{0}

and (b)

\pi_{0.5}

ID	Task	Skill
1	Place the green block in the bowl	Pick & Place
2	Place the cup on the plate	Pick & Place
3	Stack the red block on the blue block	Stacking
4	Close the drawer	Articulated Manipulation

Appendix

Table 9: Real-world robot evaluation tasks. We evaluate on four manipulation tasks spanning object placement, stacking, and articulated-object interaction.

ID	Task
1	Put both the alphabet soup and the tomato sauce in the basket
2	Put both the cream cheese box and the butter in the basket
3	Turn on the stove and put the moka pot on it
4	Put the black bowl in the bottom drawer of the cabinet and close it
5	Put the white mug on the left plate and put the yellow and white mug on the right plate
6	Pick up the book and place it in the back compartment of the caddy

Appendix

Table 10: Tasks in the LIBERO-Long benchmark. We evaluate on all 10 long-horizon manipulation tasks in the standard LIBERO-Long suite.

Category	Task	RoboCasa Skill
Articulated	Close blender lid	Open/Close Lid
	Close fridge	Open/Close Door
	Close toaster oven door	Open/Close Door
	Open cabinet	Open/Close Door
	Open drawer	Open/Close Drawer
	Open stand mixer head	Open/Close Lid

Appendix

Table 11: RoboCasa evaluation tasks. The 18 Atomic-Seen tasks span four broad interaction categories: Articulated (7), Control (4), Object-Centric (6), and Navigation (1). These categories consolidate the finer-grained atomic skills defined by RoboCasa365.

Figure 7: Measured criticality

u_{t}

vs. oracle criticality

C_{\mathrm{oracle}}

Method	Success (%)
Base	52.5
MGS	54.5
SVM	55.0
Ours	55.5

Appendix

Table 12: Success rate (%) with OpenVLA using

N=4

candidates on LIBERO-Long.

Figure 8: Chunk-level action sampling latency.

$K$	Success (%)
4	52.8
16	53.9
32	54.7
64	55.0

Appendix

Table 13: Effect of the number of samples

K

used for criticality estimation.

Table 14: Representation extraction ablation on RoboCasa with

\pi_{0.5}

. Success rate (%) under different (a) action-token positions and (b) action-expert layers.

Panel of Figure 9 — Figure 9: t-SNE visualization of action-expert representations. We visualize action-expert representations from a frozen $\pi_{0}$ policy over 500 LIBERO-Long episodes. (a) Different colors denote different tasks, revealing task-dependent structure in the representation space. (b) Successful and failed trajectories occupy distinct regions of the representation space.

Explore similar work

Verifier-free Test-Time Sampling for Vision-Language-Action Models
Oct 7, 2025Suhyeok Jang, Dongyoung Kim, Changyeon Kim +2Test-Time Scaling Vision-Language-Action Models
1KAIST · 3RLWRLD · 2Seoul National University
Flow-Support-Weighted Proposals for Test-Time Verified VLA Policies
Oct 10, 2026Chen Zhao, Zipeng Ye, Haitao Shen +3Test-Time Scaling Vision-Language-Action Models
Renmin University of China, Beijing 100872, China · Key Laboratory of Data Engineering and KnowledgeEngineering, MOE, China · Meituan, Beijing 100102, China
VeriSpace: Spatially Grounded Action Verification for Vision-Language-Action Models
Jun 9, 2026Guiyu Zhao, Longteng Guo, Junyou Zhu +6Visual Spatial Reasoning Robotic Manipulation
Institute of Automation, Chinese Academy of Sciences, Beijing, China · University of Chinese Academy of Sciences, Beijing, China

Oct 7, 2025J/K move · Enter open · S save

Verifier-free Test-Time Sampling for Vision-Language-Action Models

Suhyeok Jang, Dongyoung Kim, Changyeon Kim, Youngsuk Kim+1

1KAIST · 3RLWRLD · 2Seoul National University

Test-Time Scaling Vision-Language-Action Models VLMs for Robotics

Open paper AskPDF arXiv

Vision-Language-Action models (VLAs) have demonstrated remarkable performance in robot control. However, they remain fundamentally limited in tasks that require high precision due to their single-inference paradigm. While test-time scaling approaches using external verifiers have shown promise, they require additional training and fail to generalize to unseen conditions. We propose Masking Distribution Guided Selection (MG-Select), a novel test-time scaling framework for VLAs that leverages the model's internal properties without requiring additional training or external modules. Our approach utilizes KL divergence from a reference action token distribution as a confidence metric for selecting the optimal action from multiple candidates. We introduce a reference distribution generated by the same VLA but with randomly masked states and language conditions as inputs, providing action uncertainty while remaining aligned with the target task distribution. Additionally, we propose a joint training strategy that enables the model to learn both conditional and unconditional distributions by applying dropout to state and language conditions, thereby further improving the quality of the reference distribution. Our experiments demonstrate that MG-Select provides a reliable reference for action selection through task-relevant condition masking and consistently improves base models across diverse simulation and real-world benchmarks.

Abstract

Explore similar work

Verifier-free Test-Time Sampling for Vision-Language-Action Models

Flow-Support-Weighted Proposals for Test-Time Verified VLA Policies

VeriSpace: Spatially Grounded Action Verification for Vision-Language-Action Models