DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models

Zihao Zheng; Hangyu Cao; Chibang Tao; Zhihao Mao; Lingyue Zhang; Maoliang Li; Jiayu Chen; Hailong Zou

cs.LGMar 9, 2026

DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models

Authors: Zihao Zheng, Hangyu Cao, Chibang Tao, Zhihao Mao, Lingyue Zhang, Maoliang Li, Jiayu Chen, Hailong Zou, +4 more

Organizations: School of Computer Science, Peking University, Beijing, China. · School of Software Engineering, South China University of Technology, Guangzhou, China. · School of Artificial Intelligence, Beijing Normal University, Beijing, China. · School of Electronics Engineering and Computer Science, Peking University, Beijing China.

Abstract

Vision-language-action (VLA) models achieve favorable task performance, yet runtime errors in closed-loop execution evolve with alternating updates of actions and observations. Existing VLA quantization methods mainly trade off inference speed and model performance while rarely investigating how quantization alters runtime errors. By comparing error distributions between full-precision and quantized models, we identify distinct evolutionary patterns for the two types of errors: quantization enlarges the variance of translation error distributions and increases their dispersion, whereas the distribution center of rotation errors gradually shifts across execution steps, demonstrating cumulative drift. Motivated by this observation, we rethink the optimal quantization strategy for VLA models and propose \textit{DyQ-VLA}, a runtime-error-aware quantization framework, which dynamically selects activation precision according to execution steps and tracks as well as compensates rotation errors via accumulated quantization residuals. Corresponding operators and runtime adaptation strategies are devised within the framework to enable dynamic-precision execution. Experiments show that \textit{DyQ-VLA} achieves a 1.88 to 1.93 inference speedups while maintaining comparable or higher average task success rates. Moreover, it reduces mean execution steps by 17.7% to 19.8%. Our code is here: https://anonymous.4open.science/r/DyQ-VLA-7F51/.

Figures & tables

Figure 1: Action-Error Correlation and DyQ-VLA Overview. (a) Runtime errors and actions interaction of VLA Models. (b)Runtime-Error-Aware VLA Quantization Framework – DyQ-VLA .

Figure 2: Action–Error Correlation in Closed-Loop VLA Execution. (a) Runtime errors alter observations and subsequent actions, which in turn change runtime errors. (b) Joint distributions of correlation changes in both center and shape across steps. (c) The distribution for translation remains dispersed around a stable center, while rotation stays concentrated and shifts toward larger errors.

Figure 3: Quantization Effects and Online Tracking. (a) Quantization increases translation error variance and response sensitivity. (b) Rotation error distributions and sensitivity change only slightly. (c) Residual and feedback metrics track rotational error and translational sensitivity, respectively.

Figure 4: Framework of DyQ-VLA . Runtime feedback guides dynamic activation quantization for translation error control, while cumulative residuals guide rotation error compensation.

Method	Prec.	RoboTwin 2.0 / $\bm{\pi}_{0.5}$									LIBERO
		Clean			Randomized			Steps $\downarrow$	Speedup $\uparrow$		$\bm{\pi}_{0.5}$			GR00T-N1.5
		SR $\uparrow$	Trans.	Rot.	SR $\uparrow$	Trans.	Rot.		Step	Rollout	Step	Rollout	Mem. $\downarrow$	Step	Rollout	Mem. $\downarrow$
BF16 ref.	BF16	82.7	32.53	8.77	76.8	39.82	12.27	155.7	1.00	1.00	1.00	1.00	6.71	1.00	1.00	5.45
ActQuant	W4A16	80.9	41.65	12.10	74.2	52.38	16.48	186.4	1.11	0.93	1.10	0.92	2.63	1.09	0.94	4.03
QVLA	W4A16	80.5	42.18	10.92	73.7	51.74	15.28	188.2	1.09	0.90	1.08	0.89	2.70	1.07	0.93	4.10
QuantVLA	W4A8	81.4	40.36	15.24	74.9	50.81	20.31	181.7	1.30	1.11	1.32	1.13	3.50	1.23	1.09	4.27

Table 2: Task Performance and Inference Efficiency on RoboTwin 2.0 and LIBERO.

Figure 5: Real-World Setup and Task Examples. Details can be found in Appendix F .

Method	Prec.	Short-horizon				Multi-stage		SR (%) $\uparrow$	Task Spd. $\uparrow$
		Banana	Cup	Banana	Apple	Orange	Ordered
		Plate	Insertion	Cup	Plate	Packing	Placement
BF16 ref.	BF16	9/10	6/10	8/10	8/10	7/10	7/10	75.0	1.00
QuantVLA	W4A8	8/10	6/10	8/10	8/10	6/10	5/10	68.3	1.16
ActQuant	W4A16	8/10	6/10	8/10	8/10	6/10	5/10	68.3	1.08
QVLA	W4A16	8/10	4/10	8/10	8/10	5/10	5/10	63.3	1.04

Table 3: Real-world performance of DyQ-VLA , quantization baselines, and the BF16 reference.

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 8: Trajectory Visualizations of Runtime Error Dynamics and Their Behavioral Impact. (a) FP reference, open-loop evaluation, and closed-loop execution trajectories. (b) Identical action offsets at different steps, with trajectories projected onto the $yz$ plane. (c) Offsets of equal magnitude in different directions at the same step.

Figure 9: Action-Error Distributions Across Models and Benchmarks.

Figure 10: Tracking Rotational Runtime Error. (a) Spearman correlation with rotational runtime error. (b) Spearman correlation between signal changes and error changes. (c) NRMSE of normalized changes for error decreases $\geq 0.5^{\circ}$ ; the dashed line marks the zero-change baseline. Dots denote trajectories; boxes show medians and interquartile ranges, with $1.5\,\mathrm{IQR}$ whiskers.

Figure 12: Implementation details of the Proposed DyQ-VLA Framework.

Figure 14: Real-World Platform And End-Effector Configurations. (a) Tabletop Workspace And Hardware Components. (b) Configurations for Demonstration Collection And Task Execution.

Resource	Setting	Scope	Primary Role
OXE	Real-world	Demonstrations across multiple robot embodiments	Robot learning data
RoboCasa	Simulation	Household manipulation in kitchen environments	Task environments and demonstration data
SimplerEnv	Simulation	Manipulation setups corresponding to real robots	Policy evaluation in simulation
Our setup	Real-world	Six tabletop tasks on a single robotic platform	Task success and completion-time evaluation

Appendix

Table 6: Overview of Robotic Data And Evaluation Settings.

Task Group	Task	Objective
Short-horizon	Banana plate	Place the banana on the beige plate.
	Cup insertion	Insert one cup into the other cup.
	Banana cup	Insert the banana into the cup.
	Apple plate	Place the apple on the plate.
Multi-stage	Orange packing	Place the orange in the case and close the lid.
	Ordered placement	First place the apple on the beige plate, then place the banana on the blue plate.

Appendix

Table 7: Real-World Tasks And Objectives.

Figure 15: Representative Real-World Task Sequences. The examples show banana insertion, apple placement, and ordered placement of the apple and banana on their respective plates.

Explore similar work

Ω-QVLA: Robust Quantization for Vision-Language-Action Models via Composite Rotation and Per-step Scaling
May 27, 2026Xinyu Wang, Mingze Li, Sicheng Lyu +6Diffusion-Based Vision-Language-Actions Gradient Quantization
McGill University · Université de Montréal · Mila – Quebec AI Institute +2
VLAQuantBench: Closed-Loop Evaluation of Post-Training Quantization for Vision-Language-Action Models
Sep 21, 2026Jiuyi Xu, Qing Jin, Meida Chen +3Post-Training Quantization Closed Loop
Colorado School of Mines · Independent Researcher · University of Central Florida +2
ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models
May 19, 2026Arash Akbari, Arman Akbari, Masih Eskandar +11Post-Training Quantization Fast Inference
1Northeastern University · University of Georgia · 3Cisco Systems

May 27, 2026, cs.CVJ/K move · Enter open · S save

Ω-QVLA: Robust Quantization for Vision-Language-Action Models via Composite Rotation and Per-step Scaling

Xinyu Wang, Mingze Li, Sicheng Lyu, Dongxiu Liu+5

McGill University · Université de Montréal · Mila – Quebec AI Institute+2

Diffusion-Based Vision-Language-Actions Gradient Quantization Diffusion Policies Scaling

Open paper AskPDF arXiv

Vision-Language-Action (VLA) models unify perception, reasoning, and control within a single policy, yet their multi-billion-parameter backbones and diffusion-based action heads make on-device deployment prohibitively expensive. Prior quantization efforts offer only partial solutions, compressing the LLM backbone while leaving the DiT action head at full precision, or resorting to mixed-precision schemes, driven by the belief that uniformly quantizing the action head is inherently unstable. We challenge this assumption with Omega-QVLA, the first training-free post-training quantization framework that compresses both the language backbone and the entire diffusion action head of a VLA model to a uniform W4A4 precision, eliminating the need for mixed-precision allocation. Omega-QVLA combines a composite SVD-Hadamard rotation that equalizes per-channel weight energy while diffusing residual activation outliers with per-step DiT activation scaling quantization that absorbs dynamic-range drift across denoising steps. On LIBERO, Omega-QVLA compresses Pi 0.5 and GR00T N1.5 to W4A4 with 98.0% and 87.8% task success rates, matching or exceeding their FP16 references of 97.1% and 87.0%, while reducing the static memory footprint by 71.3%. Real-world manipulation experiments further confirm smooth, accurate manipulation where prior methods fail. Code is available at https://github.com/UCMP13753/Omega-QVLA.

Abstract

Explore similar work

Ω-QVLA: Robust Quantization for Vision-Language-Action Models via Composite Rotation and Per-step Scaling

VLAQuantBench: Closed-Loop Evaluation of Post-Training Quantization for Vision-Language-Action Models

ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models