Skip to content
Preprint

D$^2$-VLA: Dual-Memory Dual-Frequency Vision-Language-Action Model For Long Dynamic Manipulation

Sep 2026 · 0 citations · 33 references
Computer Science

TL;DR

D$^2$-VLA is presented, which combines dual memory and dual-frequency control at the KV-cache interface of a pretrained VLA, which uses block-wise causal KV caching to encode observations incrementally and constructs separate historical KV read views for the VLM and action expert.

Abstract

Long-horizon manipulation requires robots to remember cues that are no longer in view while responding to moving objects. Yet vision-language-action (VLA) policies often rely on the latest observation, and refreshing their visual context typically requires another costly vision-language model (VLM) pass. We present D$^2$-VLA, which combines dual memory and dual-frequency control at the KV-cache interface of a pretrained VLA. D$^2$-VLA uses block-wise causal KV caching to encode observations incrementally and, guided by distinct temporal attention patterns, constructs separate historical KV read views for the VLM and action expert. Between periodic VLM updates, a gated adapter incorporates fresh visual features into the latest history-conditioned KV block, while a short fast-memory queue supports action replanning. We introduce DOMINO-Long, a ten-task benchmark requiring robots to use earlier visual cues when manipulating moving objects. D$^2$-VLA achieves complete-task success rates of 29.3\% on DOMINO, compared with 9.6\% for $\pi_{0.5}$ and 17.2\% for PUMA, and 60.0\% on DOMINO-Long, compared with 35.4\% and 20.6\%, respectively. It improves success rates on eight real-robot tasks and reaches 97.5\% on LIBERO-Long and 74.3\% on RoboTwin 2.0.

View source

Similar papers

Preprint Sep 2026

Remember What You Did: Action-History Memory with Dual-Expert Denoising for Long-Horizon Vision-Language-Action Policies

Vision-language-action (VLA) models have driven rapid progress in robotic manipulation, demonstrating strong fine-grained control and promising performance on long-horizon tasks. However, many existing VLAs lack explicit access to interaction history, making them vulnerable to perceptual aliasing: similar current obser...

Ya-Xin Zhao, Dian-Ye Huang, Chen-Wei Wang et al. · 0 citations
Preprint Aug 2026

Skills in Weights, Memory in Code: Hybrid Learning for Memory-Dependent Robot Manipulation

HyMeS, a hybrid learning framework that leverages the reasoning and memory-management capabilities of coding agents to steer a Markovian VLA for memory-dependent manipulation, is proposed, enabling data-efficient compositional generalization.

Yun-Hao Zhao, Zhen-Yang Ni, Haoyang Chen et al. · 0 citations
Sep 2026

FIVE-VLA: Fast and EffectIVE Autonomous Driving with Recurrent Action Memory

Recurrent Action Memory (RAM) is proposed, a lightweight module that conditions action prediction on previous action tokens, providing temporal context critical for manoeuvres such as overtaking and emergency braking.

Kemal Oksuz, Alexandru Buburuzan, Yu-Han Yao et al. · 0 citations
Sep 2026

C2Nav: Compare Before You Commit for Zero-Shot Vision-and-Language Navigation

This work instantiates a complementary model-robot interface: the VLM compares controller-constructed alternatives, while geometry, thresholds, action magnitude, and execution remain on the physical side in C2Nav, a training-free framework with three coordinated faculties.

Run-Tian Zheng, Cong-Peng Zhang, Ying Liu · 1 citation
Preprint Sep 2026

T$^2$Mem: Learning Test-Time Memory for Robotics

Memory-dependent robotic manipulation requires policies to use information that is no longer available in the current observation. Retaining history alone is insufficient: memory must preserve information that supports future actions. One challenge is whether a memory-free foundation model can learn to retain and use h...

Yi-Ze Liu, Huang Huang, Yi-Ning Hong et al. · 2 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.