Recurrent-Depth VLA: Implicit Test-Time Compute Scaling of Vision-Language-Action Models via Latent Iterative Reasoning

• URL: http://arxiv.org/abs/2602.07845v1
• Date: Sun, 08 Feb 2026 07:21:01 GMT
• Title: Recurrent-Depth VLA: Implicit Test-Time Compute Scaling of Vision-Language-Action Models via Latent Iterative Reasoning
• Abstract summary: Recurrent-Depth VLA (RD-VLA) is an architecture that achieves computational adaptivity via latent iterative refinement rather than explicit token generation.<n>RD-VLA employs a recurrent, weight-tied action head that supports arbitrary inference depth with a constant memory footprint.
• Score: 38.78160379823724
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Current Vision-Language-Action (VLA) models rely on fixed computational depth, expending the same amount of compute on simple adjustments and complex multi-step manipulation. While Chain-of-Thought (CoT) prompting enables variable computation, it scales memory linearly and is ill-suited for continuous action spaces. We introduce Recurrent-Depth VLA (RD-VLA), an architecture that achieves computational adaptivity via latent iterative refinement rather than explicit token generation. RD-VLA employs a recurrent, weight-tied action head that supports arbitrary inference depth with a constant memory footprint. The model is trained using truncated backpropagation through time (TBPTT) to efficiently supervise the refinement process. At inference, RD-VLA dynamically allocates compute using an adaptive stopping criterion based on latent convergence. Experiments on challenging manipulation tasks show that recurrent depth is critical: tasks that fail entirely (0 percent success) with single-iteration inference exceed 90 percent success with four iterations, while simpler tasks saturate rapidly. RD-VLA provides a scalable path to test-time compute in robotics, replacing token-based reasoning with latent reasoning to achieve constant memory usage and up to 80x inference speedup over prior reasoning-based VLA models. Project page: https://rd-vla.github.io/

Related papers

• $τ_0$-VLA: a Hierarchical Robot Foundation Model with World-Model-Guided Test-Time Computation [51.05516457377339]
  Long-horizon robot manipulation requires a robot to both execute individual skills reliably and sequence them coherently over extended tasks.<n>We introduce $_0$-VLA, a hierarchical robot foundation model that formulates high-level subtask generation as a compute-scalable inference problem.<n>A low-level policy then executes the generated subtask across multiple robot embodiments.
  arXiv  Detail & Related papers  (2026-08-17T17:59:11Z)
• LoopVLA: Learning Sufficiency in Recurrent Refinement for Vision-Language-Action Models [13.30873593845724]
  LoopVLA is a recurrent Vision-Language-Action architecture that learns representation refinement, action prediction, and sufficiency estimation.<n>We show that LoopVLA pushes the efficiency-performance frontier of VLA policies, reducing parameters by 45% and improving inference throughput by up to 1.7 times.
  arXiv  Detail & Related papers  (2026-05-11T03:51:22Z)
• LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models [35.78769938627291]
  We introduce textbfLooped Depth Up-Scaling (LoopUS), a post-training framework that converts a standard pretrained LLM into a looped architecture.<n>LoopUS recasts the pretrained LLM into an encoder, a looped reasoning block, and a decoder.<n>These mechanisms transform a standard non-looped model into a looped form while stabilizing it against both computational bottlenecks and representation collapse.
  arXiv  Detail & Related papers  (2026-05-10T11:05:20Z)
• VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Model [54.35791816657227]
  We introduce textbfVLA-ATTC', a framework that endows VLA models with adaptive test-time compute.<n>VLA-ATTC employs an uncertainty-based cognitive clutch'' to dynamically transition from reflexive execution to a TTC deliberation phase.<n>On the LIBERO-LONG benchmark, VLA-ATTC reduces the failure rate of the SOTA model PI0.5 by over 50%.
  arXiv  Detail & Related papers  (2026-05-02T02:13:11Z)
• Sentinel-VLA: A Metacognitive VLA Model with Active Status Monitoring for Dynamic Reasoning and Error Recovery [62.75419724651416]
  We introduce textbfSentinel-VLA, a metacognitive VLA model equipped with an active sentinel'' module to monitor real-time execution status.<n>All training data is automatically generated and annotated through our designed pipeline.<n>Real-world experiments demonstrate that Sentinel-VLA boosts the task success rate by over 30% compared to the SOTA model, PI0.
  arXiv  Detail & Related papers  (2026-05-02T02:10:54Z)
• DFM-VLA: Iterative Action Refinement for Robot Manipulation via Discrete Flow Matching [20.252867273996085]
  We propose DFM-VLA, a discrete flow matching VLA for iterative refinement of action tokens.<n>DFM-VLA consistently outperforms strong autoregressive, discrete diffusion, and continuous diffusion baselines in manipulation performance.
  arXiv  Detail & Related papers  (2026-03-27T11:38:43Z)
• RaBiT: Residual-Aware Binarization Training for Accurate and Efficient LLMs [5.782015253162346]
  Residual binarization enables matmul-free inference by stacking binary layers.<n>We propose RaBiT, a novel quantization framework that resolves coadaptation by algorithmically enforcing a residual hierarchy.<n>RaBiT achieves state-of-the-art performance, rivals even hardware-intensive Vector Quantization (VQ) methods, and delivers a $4.49times$ inference speed-up over full-precision models.
  arXiv  Detail & Related papers  (2026-02-05T06:41:11Z)
• Learning to Accelerate Vision-Language-Action Models through Adaptive Visual Token Caching [25.295588774254952]
  We reformulate inference acceleration as a learnable policy optimization problem.<n>We propose a novel framework that integrates a task-aware decision-making process directly into the Vision-Language-Action model.<n>Our method achieves a 1.76x wall-clock inference speedup while simultaneously improving the average success rate.
  arXiv  Detail & Related papers  (2026-01-31T12:12:51Z)
• FASTer: Toward Efficient Autoregressive Vision Language Action Modeling via Neural Action Tokenization [61.10456021136654]
  We introduce FASTer, a unified framework for efficient and general robot learning.<n>FASTerVQ encodes action chunks as single-channel images, capturing global-temporal dependencies while maintaining a high compression ratio.<n>FASTerVLA builds on this tokenizer with block-wise autoregressive decoding and a lightweight action expert, achieving both faster inference and higher task performance.
  arXiv  Detail & Related papers  (2025-12-04T16:21:38Z)
• Steering Vision-Language-Action Models as Anti-Exploration: A Test-Time Scaling Approach [78.4812458793128]
  We propose textbfTACO, a test-time-scaling framework that applies a lightweight pseudo-count estimator as a high-fidelity verifier of action chunks.<n>Our method resembles the classical anti-exploration principle in offline reinforcement learning (RL), and being gradient-free, it incurs significant computational benefits.
  arXiv  Detail & Related papers  (2025-12-02T14:42:54Z)
• MesaNet: Sequence Modeling by Locally Optimal Test-Time Training [67.45211108321203]
  We introduce a numerically stable, chunkwise parallelizable version of the recently proposed Mesa layer.<n>We show that optimal test-time training enables reaching lower language modeling perplexity and higher downstream benchmark performance than previous RNNs.
  arXiv  Detail & Related papers  (2025-06-05T16:50:23Z)
• TANTE: Time-Adaptive Operator Learning via Neural Taylor Expansion [9.639856161665962]
  We propose a novel operator-learning framework that produces continuous-time predictions with adaptive step sizes.<n>TANTE predicts future states by performing a Taylor expansion at the current state, where neural networks learn both the higher-order temporal derivatives and the local radius of convergence.<n>We demonstrate the effectiveness of TANTE across a wide range of PDE benchmarks, achieving superior accuracy and adaptability compared to fixed-step baselines.
  arXiv  Detail & Related papers  (2025-02-12T17:09:13Z)
• COrAL: Order-Agnostic Language Modeling for Efficient Iterative Refinement [80.18490952057125]
  Iterative refinement has emerged as an effective paradigm for enhancing the capabilities of large language models (LLMs) on complex tasks. We propose Context-Wise Order-Agnostic Language Modeling (COrAL) to overcome these challenges. Our approach models multiple token dependencies within manageable context windows, enabling the model to perform iterative refinement internally.
  arXiv  Detail & Related papers  (2024-10-12T23:56:19Z)
• ConStruct-VL: Data-Free Continual Structured VL Concepts Learning [57.86651057895222]
  We introduce the first Continual Data-Free Structured VL Concepts Learning (ConStruct-VL) benchmark. We propose a data-free method comprised of a new approach of Adrial Pseudo-Replay (APR) which generates adversarial reminders of past tasks from past task models. We show this approach outperforms all data-free methods by as much as 7% while even matching some levels of experience-replay.
  arXiv  Detail & Related papers  (2022-11-17T18:57:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.