ProbeFlow: Training-Free Adaptive Flow Matching for Vision-Language-Action Models

Zhou Fang; Jiaqi Wang; Wenzhuo Li; Yi Zhou; Qiongfeng Shi

cs.ROMar 18, 2026

ProbeFlow: Training-Free Adaptive Flow Matching for Vision-Language-Action Models

Authors: Zhou Fang, Jiaqi Wang, Wenzhuo Li, Yi Zhou, Qiongfeng Shi

Organizations: School of Computer Science and Engineering, Southeast University, China · Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications, Ministry of Education, China · School of Electronic Science & Engineering, Southeast University, China

Abstract

Vision-Language-Action (VLA) policies increasingly use Flow Matching (FM) action heads to generate action chunks as continuous trajectories. At inference, an FM head solves an ordinary differential equation (ODE): starting from a prior action latent, it repeatedly queries a learned velocity field to advance the state toward the decoded action chunk. Because each query depends on the preceding state, this iterative decoding can impose substantial sequential action-head cost at deployment. We propose ProbeFlow, a training-free framework that routes each action chunk to an integration schedule using a Lookahead Consistency Probe. Optional oscillation-conditioned (OSC) threshold adaptation reuses the same velocity evaluations and can trade part of the acceleration for potentially better task performance. On MetaWorld, the fixed-threshold configuration uses 2.60 scheduled steps (2.92 action-head evaluations) and reports 15.9 ms action-head latency, versus 235.7 ms for 50-step Euler, while obtaining 83.2% success versus 82.5%. On LIBERO, it uses 4.53 scheduled steps (5.09 action-head evaluations), reports 32.7 ms, and obtains 88.7% success, compared with 278.7 ms and 92.5% for 50-step Euler. We also evaluate OSC on three real-world manipulation tasks with 50 trials per method and task. These evaluations characterize action-decoding compute-performance trade-offs without additional training.

Figures & tables

Fig. 1: Overview of the proposed ProbeFlow framework. Left: A conditional action flow is integrated from a prior state to an action chunk. Middle: The Lookahead Consistency Probe compares initial and lookahead velocity directions through cosine similarity $\mathcal{S}$ , which the scheduler maps to an adaptive step count $N$ under a deployment latency budget; optional OSC maps a sign-change ratio to an effective threshold. Right: High directional consistency ( $\mathcal{S}\approx 1$ ) selects a coarse schedule that reuses the two probe evaluations; lower consistency selects a denser schedule.

Method	Reported Count	Flow Solver (ms) $\downarrow$	Success Rate (%) $\uparrow$
Fixed-Euler ( $N=50$ )	50.0	235.7	82.5
Fixed-Euler ( $N=20$ )	20.0	99.2	80.6
Fixed-Euler ( $N=10$ )	10.0	53.4	81.6
Fixed-Euler ( $N=3$ )	3.0	23.7	72.4
Fixed-Euler ( $N=2$ )	2.0	13.2	68.5
RK45	68.9	2823.8	63.0

TABLE I: Action-head latency and success rate on MetaWorld. Fixed-threshold and baseline rows aggregate 5 seeds; the OSC row aggregates 3 seeds. For ProbeFlow rows, the parenthesized value is the measured average NFE; all probe evaluations are included in action-head latency. Visual encoding is excluded because ProbeFlow does not modify the visual encoder.

Method	Reported Count	Flow Solver (ms) $\downarrow$	Success Rate (%) $\uparrow$
Fixed-Euler ( $N=50$ )	50.0	278.7	92.5
Fixed-Euler ( $N=20$ )	20.0	109.3	92.3
Fixed-Euler ( $N=10$ )	10.0	54.5	89.0
Fixed-Euler ( $N=5$ )	5.0	34.0	86.9
Fixed-Euler ( $N=4$ )	4.0	26.1	83.2
RK45	101.2	3769.3	91.2

TABLE II: Action-head latency and success rate on LIBERO. Fixed-threshold and baseline rows aggregate 5 seeds; the OSC row aggregates 3 seeds. For ProbeFlow rows, the parenthesized value is the measured average NFE; all probe evaluations are included in action-head latency. Visual encoding is excluded because ProbeFlow does not modify the visual encoder.

Fig. 2: Task-level directional consistency and scheduled integration counts on MetaWorld MT50 at

\Delta t_{\mathrm{probe}}=0.5

\epsilon=0.008

, with OSC disabled. Each point represents one task, averaged equally over five seeds with ten episodes per task and seed. Error bars denote standard deviations across seed-level means.

Panel of Figure 5 — Fig. 3: Qualitative examples of adaptive step scheduling for two MetaWorld rollouts. Top: In the shown Basketball rollout, the selected schedules contain more steps (red) around object interaction and insertion. Bottom: In the shown Button Press rollout, the selected schedules often use the minimum step count (blue) during transit. Colors visualize the selected schedules.

Model	Benchmark	Fixed Steps	Fixed Success	OSC Success	$\Delta$ Success	OSC Avg. $N$
Evo-1	MetaWorld	10	81.6%	84.0%	+2.4 pp	3.84
Evo-1	LIBERO	10	89.0%	89.25%	+0.25 pp	5.55
OpenPI / PI0.5	LIBERO-90	10	27.97%	28.26%	+0.29 pp	2.32
OpenPI / PI0.5	LIBERO-object	10	100.0%	99.0%	-1.0 pp	2.02
OpenPI / PI0.5	LIBERO-goal	10	97.0%	96.33%	-0.67 pp	2.02
OpenPI / PI0.5	LIBERO-10	10	91.0%	94.0%	+3.0 pp	2.04

TABLE III: Fixed-step reference versus OSC with

r_{\mathrm{ref}}=0.02

. OSC results aggregate 3 seeds; the Evo-1 fixed-step references are the 5-seed results from Tables I and II . Differences are descriptive comparisons, not paired estimates of the effect of enabling OSC. ‘–’ denotes an unavailable measurement. ‘OSC Avg. Integration

N

’ is reported only when the implementation exposes an adaptive integration schedule; fixed denoising-loop lengths are not relabeled as adaptive steps. For Evo-1,

N

is a scheduled integration count and the corresponding action-head NFE follows the convention stated in Sec. IV-B . The OpenPI/PI0.5 LIBERO-90 result uses 900 episodes per seed; suite-level results use the corresponding evaluation episodes.

Domain	Threshold $\epsilon$	Avg. Scheduled $N$	Action Head (ms) $\downarrow$	Success Rate (%)
MetaWorld	0.001	15.7	90.6	82.8
MetaWorld	0.002	8.6	47.3	81.1
MetaWorld	0.004	4.6	28.4	82.7
MetaWorld	0.008	2.6	15.9	83.2
LIBERO	0.001	19.7	135.1	91.6
LIBERO	0.002	14.1	104.4	92.0

TABLE IV: Sensitivity analysis of the directional-deviation tolerance

\epsilon

across MetaWorld and LIBERO, aggregated over 5 seeds with 10 episodes per task and seed. ‘Avg. Scheduled

N

’ is the selected integration schedule; measured action-head latency includes the probe evaluation.

Task	Method	Success	95% Wilson CI	Solver (ms)	Steps	Obs. $\Delta$ (normalized)	Obs. jerk (normalized)
pick-place fruit	$\pi_{0.5}$	82%	69.2–90.2%	183.9	10.0	0.06	0.03
pick-place fruit	$\pi_{0.5}$	84%	71.5–91.7%	354.7	20.0	0.055	0.025
pick-place fruit	$\pi_{0.5}$	84%	71.5–91.7%	885.0	50.0	0.050	0.020
pick-place fruit	ProbeFlow OSC	90%	78.6–95.7%	59.6	$\mathbf{2.6}$	0.07	0.04
flip mug	$\pi_{0.5}$	52%	38.5–65.2%	185.3	10.0	0.04	0.03
flip mug	$\pi_{0.5}$	54%	40.4–67.0%	357.0	20.0	0.035	0.025

TABLE V: Multi-task real-world evaluation with 50 trials per method and task. Success rates and 95% Wilson intervals are computed from the number of successful trials out of 50. The reported steps are averaged across episodes.

Fig. 4: Probe-horizon ablation on MetaWorld (MT50). Points show means over five seeds; error bars denote standard deviations. The average scheduled integration count (dashed orange) generally increases with the probe horizon, while success-rate means (solid blue) range from 83.2% to 84.0% for horizons from 0.5 to 0.8.

Panel of Figure 10 — Fig. 5: Representative real-world manipulation sequences. Each row shows the temporal progression of one task under ProbeFlow .

Explore similar work

AdaVLA: Adaptive Step Flow Matching for Training-free Acceleration of Vision-Language-Action Models
Aug 29, 2026Sunghwan Han, Youngtae Han, Youngmin YiFlow Matching Vision-Language-Action Models
Dept. of Artificial Intelligence, Sogang University, Seoul, 04107, Republic of Korea
FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference
Aug 27, 2026Zekai Li, Jiaming Tang, Zhijian LiuEfficient VLA Model Inference VLMs for Robotics
UC San Diego · MIT
RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Models
Sep 28, 2026Yuhan Chen, Ke Yu, Pengfei Liu +3Efficient VLA Model Inference Vision-Language-Action Models
Zhejiang University · Nanyang Technological University

Aug 29, 2026J/K move · Enter open · S save

AdaVLA: Adaptive Step Flow Matching for Training-free Acceleration of Vision-Language-Action Models

Sunghwan Han, Youngtae Han, Youngmin Yi

Dept. of Artificial Intelligence, Sogang University, Seoul, 04107, Republic of Korea

Flow Matching Vision-Language-Action Models VLMs for Robotics

Open paper AskPDF arXiv

Vision-Language-Action (VLA) models, built upon Vision-Language Models (VLMs), have significantly enhanced robotic capabilities by leveraging internet-scale knowledge and multimodal reasoning. However, the intensive computational overhead of VLAs constrains on-device deployment, hindering real-time responses to environmental changes. While various acceleration techniques have been proposed, they often rely on fine-tuning or access to training datasets, which are frequently unavailable due to privacy and proprietary concerns. Moreover, although flow-matching-based VLAs have emerged as efficient alternatives to standard diffusion models, current acceleration efforts largely target VLM inference costs, failing to address the iterative ODE solving process inherent in flow matching inference. To address these limitations, we propose AdaVLA, an online, training-free adaptive framework for fast yet accurate flow-matching-based Vision-Language-Action models. We introduce a novel metric derived from the flow matching trajectory curvature to quantify action generation confidence during inference. This metric enables the dynamic reduction of inference steps and the adaptive adjustment of MLP pruning ratios through an efficiently computed importance evaluation, requiring no access to training data. Experimental results on the LIBERO benchmark using a Jetson AGX Orin device demonstrate that our method achieves

1.87\times

and

2.24\times

speedups for

π_{0.5}

and X-VLA, respectively, with negligible degradation in success rates. Furthermore, we validate the robustness of our approach on real-world robotic tasks using SmolVLA.

Abstract

Explore similar work

AdaVLA: Adaptive Step Flow Matching for Training-free Acceleration of Vision-Language-Action Models

FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference

RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Models