Efficient VLA Model Inference

Flow-based vision-language-action (VLA) models generate action chunks for temporally coherent robot motion, but chunked control creates a fundamental closed-loop trade-off: long chunks provide smooth execution, whereas frequent replanning improves reactivity at the cost of action discontinuities. We introduce REACT, a rolling-denoising framework that makes flow-based VLAs more reactive while preserving long-horizon context. Instead of regenerating entire action chunks from scratch, REACT maintains a persistent action buffer with staggered flow timesteps. At each control step, the full horizon is denoised using the latest observation, the cleanest action block is executed, partially refined future blocks are shifted forward, and fresh noise is appended to the tail. As a result, each executed action block is refined across multiple recent observations before deployment. To support real-time control, we further introduce dual decoupling, which separates sensing, VLM encoding, DiT denoising, and action execution, enabling high-frequency observation updates and action streaming under practical compute constraints. Across the RoboTwin 2.0 simulation benchmark and real-world tasks spanning bimanual manipulation and dynamic control on multiple robot platforms, REACT improves task success and reduces reaction latency while producing smoother trajectories than frequent-replanning and asynchronous baselines.

JEV models make multimodal decisions by directly scoring candidates. Although the common context is encoded once, candidate evaluation can still repeat matching token histories, duplicate inference states, and execute the full backbone. In this paper, we study these sources of redundancy and present FastJEV for compact candidate evaluation. We jointly organize history reuse and state storage, since sharing computation requires preserving states for later branches. We first introduce shared context anchoring to reuse recurrent initial states and omit unused final recurrent caches. We extend this reuse through candidate prefix sharing, retaining the intermediate states needed by subsequent branches. To further reduce the depth of these paths, we apply decision guided pruning based on relative score changes measured on a small unlabeled set. Our method retains full context encoding and all candidates without additional training. We evaluate FastJEV across three OmniJev model sizes on five public benchmarks and reconstructed LIBERO-10 offline questions. At the selected pruning budgets, the complete method reduces candidate depth by 43.75% to 45.83%, while retaining 93.66% to 97.52% of the original task scores on average across the six evaluation sets. Through controlled experiments, we show how candidate overlap and branching structure affect the execution cost of history reuse. In our implementation, candidate prefix sharing can reduce repeated computation while increasing latency. These findings motivate designing sharing granularity and execution schedules together for efficient JEV inference.

ANN-to-SNN conversion offers a practical route toward energy-efficient spiking Vision-Language-Action (VLA) models by bypassing the substantial cost of training large-scale SNNs from scratch. However, existing methods often require many timesteps to maintain competitive performance, resulting in substantial inference latency for real-time VLA deployment. To address this challenge, we introduce SpikingVLA, an ANN-to-SNN conversion framework that enables accurate and low-latency spiking VLA inference. Specifically, we propose a Dendritic Integrate-and-Fire (DIF) neuron that alleviates channel-wise activation outliers through dendritic mixing and adaptive somatic firing, enabling accurate ANN-to-SNN conversion with fewer timesteps. Building on DIF neurons, we further introduce an asynchronous execution mechanism that overlaps temporal computation across VLA components, reducing synchronization overhead and latency. Extensive experiments demonstrate that SpikingVLA achieves competitive navigation performance with substantially improved inference efficiency. Compared with existing spiking VLA methods, SpikingVLA improves SR and SPL by 11.9% and 12.6%, respectively, while reducing first-action latency by 11.2$\times$. These results establish SpikingVLA as a practical framework for deploying pretrained VLA models with high-performance and low-latency spiking inference.

While vision-language-action (VLA) models have advanced rapidly, running them at every control step remains expensive. Prior work accelerates VLA inference using techniques like action chunking and visual-token pruning, typically evaluating based on latency and average task success. However, acceleration may discard information and break tasks the original policy would solve, a risk hidden by average metrics. Measuring these failures is challenging because action deviations compound over closed-loop trajectories, meaning task failure is only observable across full episodes. We therefore define an acceleration-induced failure via paired rollouts from identical initial conditions, tracking when the reference succeeds but the accelerated policy fails. To manage this, we introduce CARE, an approach for certified accelerator selection. CARE uses paired rollouts on a calibration set to provide finite-sample guarantees that acceleration-induced failure risk stays below a user-specified budget. It deploys the fastest certified candidate, falling back to the reference if none qualify. By relying only on terminal outcomes and measured compute, CARE applies unchanged across diverse acceleration mechanisms, while sequential testing and failure-triggered reference rollouts keep certification affordable. On four LIBERO suites with OpenVLA-OFT, CARE certifies $9.0$--$10.8\times$ speedups while guaranteeing (at $95\%$ confidence) that at least $85.8\%$ of reference-solved episodes are preserved. Under tight budgets, selectors without guarantees exceed the budget in up to $75\%$ of trials, whereas CARE stays within budget and its sequential form uses $78.9\%$ fewer rollouts than exhaustive evaluation. CARE further generalizes to flow-step reduction for $π_{0.5}$, and to Qwen3.5-9B and Llama-3.1-8B agents in Crafter.

Vision-language-action (VLA) policies repeatedly invoke inference to control robots, making graphics processing unit (GPU) energy a recurring cost of task execution. Reducing energy per inference call, however, may not reduce energy per successful task if numerical errors increase failures or slower inference prolongs execution. We therefore target GPU energy per successful task while preserving task success and keeping the inference-latency increase within 10%. Our approach builds on two observations: quantization sensitivity varies across action classes, model layers, and weights versus activations; and numerical precision changes the workload, shifting favorable GPU operating points. We introduce ActTune, an action-aware framework that connects layer-wise precision allocation with workload-dependent GPU operating-point selection over requested frequency--power-cap pairs. A lightweight decision tree learns its splits and leaf precision configurations directly from configuration action errors, then selects precision before each policy call. The controller forecasts the next workload and applies the selected GPU operating point asynchronously using a lookup table calibrated under a latency budget. A shared resident quantized weight bank enables configuration switching without weight reconstruction or additional policy evaluations. On LIBERO, a benchmark for lifelong robot learning, ActTune improves mean task success by up to 2.3% relative to state of the art. Relative to the original BF16 implementations, it delivers up to $2.02\times$ faster inference and, with GPU operating-point adaptation, reduces energy per successful task by up to 76.8%.

Vision-language-action (VLA) models increasingly rely on diffusion- or flow-matching-based action heads to generate continuous robot actions. These action heads typically process the denoising trajectory in a largely uniform manner. However, we observe that the conditioning focus naturally shifts across denoising stages: early stages combine language instructions and visual observations to establish a coarse action trajectory, whereas later stages place greater emphasis on current visual observations for action alignment. Based on this insight, we introduce StairVLA, a stage-aware hierarchical action generation framework that uses partially denoised actions as a natural interface between coarse long-horizon action generation and local refinement. A high-level VLA performs early denoising to produce a reusable long-horizon partially denoised action trajectory, while a lightweight refiner operates at a higher frequency to refine local action chunks using the latest observations. This design amortizes expensive high-level VLA computation while preserving frequent closed-loop correction. On LIBERO, our GR00T-style instantiation improves average success from 96.5% to 97.8% while reducing amortized inference latency from 115.0 ms to 44.2 ms per action chunk. More broadly, across two VLA backbones, simulation benchmarks, and real-robot tasks, StairVLA consistently reduces inference cost while maintaining strong task performance.

Generative action models based on diffusion and flow matching have been increasingly adopted in vision-language-action (VLA) policies for their ability to capture diverse behaviors, including multiple valid action sequences under the same observation and instruction. Their iterative sampling procedures, however, require repeated network evaluations to generate each action chunk, increasing inference latency in closed-loop control. We propose ESP (Energy-Score Policy), a teacher-free approach that maps policy context and noise directly to an action chunk in a single network evaluation. ESP trains the action head with the energy score rather than mean squared error. Whereas squared-error regression targets the conditional mean, the energy score is strictly proper: its expected value is uniquely minimized by the target distribution. This provides a principled objective for learning multimodal action distributions without iterative sampling, with exact recovery at the population optimum when the model can represent the target distribution. Experiments on both simulation and real-world manipulation tasks demonstrate competitive task success with substantially lower action-generation latency than the flow matching baseline. These results support direct distributional learning as an efficient alternative to iterative generative robot policies.

Adapting vision-language-action (VLA) models to deployment-time distribution shifts is important for reliable robotic operation, but conventional first-order adaptation can exceed the memory budget of inference-oriented deployment platforms. Zeroth-order (ZO) optimization offers a forward-only alternative with inference-level memory, but accurate gradient estimation requires many perturbation queries, making naive ZO prohibitively slow for large VLA models. We present VLA-ZO, a framework for fast ZO adaptation that exploits the structure of VLA computation. By confining adaptation to the action side, VLA-ZO keeps the expensive vision-language prefix frozen and reuses its conditioning states across perturbation queries and optimizer steps, while schedule-aware prefetching hides state-transfer overhead. On LIBERO camera-viewpoint shifts, VLA-ZO reduces end-to-end adaptation time by 25.59$\times$ at $q=16$ and 32.54$\times$ at $q=64$ relative to baseline ZO, while improving average task success from 48.27% without adaptation to 58.17% and 63.58%, respectively. These results show that making ZO faster can make larger query budgets practical, providing a promising path toward resource-efficient VLA adaptation on deployment platforms.

Vision-Language-Action (VLA) models serve as unified policies for robotic manipulation, yet their expensive inference forces robots to pause between policy calls, resulting in stop-and-go execution that interrupts smooth motion and prolongs task completion. Extending the action chunk reduces policy calls and hence these pauses, but predicting farther into the future makes long-chunk execution unreliable. To understand where this unreliability arises, we analyze action errors within long chunks and find that they concentrate around transitions between manipulation subskills, growing sharply with chunk length. This suggests the importance of transition timing, i.e., when to switch subskills within a chunk. Motivated by this observation, we introduce RACE (Reliable Action-Chunk Extension), a framework that predicts the transition timing from an auxiliary one-step denoising pass and conditions action generation on it. By learning and conditioning on transition timing, RACE reduces errors at subskill transitions and enables reliable execution of longer action chunks. Across simulation benchmarks, RACE outperforms fine-tuning at the same chunk length; with 2x longer chunks, it surpasses recent state-of-the-art and efficient VLAs in success rate, and with 4x longer chunks, it remains competitive. On a real robot, RACE uses 4x longer chunks, which reduces the idle time caused by stop-and-go execution by about 5x, while achieving a higher success rate than fine-tuning with the same chunk length. Code and a real-robot demo are available at https://github.com/Seonghoon-Yu/RACE-VLA

Visual token pruning is an effective way to accelerate vision-language models and is especially useful for vision-language-action (VLA) inference, where many visual tokens must be processed before predicting robot actions. Existing pruning methods usually estimate which tokens can be pruned based on attention scores or feature diversity, retaining tokens that are either highly attended or visually different from others. However, most of them use fixed pruning schedules, such as pruning once at a preset layer or pruning at uniformly spaced layers. Such schedules can be risky for VLA models, because the model may not know which visual regions matter for the action in early layers. Tokens that look unimportant at first may become useful after the model combines visual observations with the language instruction. In this work, we propose SAPrune, a training-free visual token pruning framework for efficient VLA inference. Instead of pruning at fixed layers, SAPrune uses a small calibration set to observe how action-to-visual attention changes across layers, and chooses pruning layers only after the attention pattern becomes more reliable. At each selected layer, SAPrune applies a dual-path pruning rule: one path protects strongly attended visual tokens from pruning, while the other prevents useful surrounding context from being discarded. Experiments on LIBERO, SIMPLER, and real-world robotic tasks show that SAPrune prunes 87.5% of visual tokens and achieves up to 1.718x inference speedup while maintaining competitive task success rates.

Vision-language-action (VLA) models translate multimodal observations into low-level robot actions. During robot operation, each control period sets an inference deadline, and overruns leave the robot acting on stale observations, reducing task success. Meeting this deadline motivates on-device or nearby edge execution, where a single robot requires batch-1 inference outside the design point of LLM serving systems. Although VLA architectures combine familiar vision-language, autoregressive, and diffusion-style components, their runtime behavior in this batch-1 control setting remains uncharacterized. We characterize four representative VLA models on an edge GPU server and two onboard SoCs, using single-inference profiling and 43,200 closed-loop episodes. Action tensor dimensionality determines whether a stage is memory- or compute-bound, platform balance can shift that bottleneck, and GPU frequency scaling yields a platform-dependent energy-latency sweet spot. In closed-loop operation, overlapping inference with action execution creates an accuracy-speed-energy tradeoff, and no configuration is Pareto-dominant across deployment SLOs. These results guide joint design of VLA model architectures, hardware, and runtime policies.

Dual-system Vision-Language-Action (VLA) models improve real-time robotic control by pairing a slow, reasoning-capable generalist with a fast specialist action expert. However, existing methods invoke the generalist at a fixed frequency, ignoring the fact that decision-making complexity varies throughout a rollout. This static strategy wastes computation in easy phases and can delay renewed reasoning when the scene changes unexpectedly. We propose TUD (Triggering generalist reasoning via predictive Uncertainty for Dual-system VLA), an adaptive inference framework that selectively skips unnecessary generalist calls. TUD measures the cross-step dispersion of action re-predictions at the upcoming chunk slot under the cached generalist context, as a predictive uncertainty signal. This signal captures how much the future action plan shifts as new observations arrive and is computed from forwards the architecture already runs, requiring neither manual phase labels nor an auxiliary uncertainty model. On VLA-Arena, it achieves a higher success rate at matched call budgets than alternative uncertainty baselines while maintaining low wall-clock overhead, and more consistently separates successful from failed rollouts. Also, TUD finds a more favorable cost-success trade-off than non-adaptive baselines, tracing an entire operating curve as a single threshold is varied, and substantially reduces VLM calls at matched success rate. The same trade-off appears in our real-robot experiments, where TUD cuts generalist calls by 75% relative to the strongest fixed-interval baseline while achieving an even higher success rate. Our results suggest that predictive uncertainty provides a practical criterion for adaptive reasoning in efficient VLA control.

Vision-language-action (VLA) models increasingly incorporate intermediate reasoning to improve robotic manipulation, yet existing approaches primarily reason about observed states without explicitly anticipating future scene evolution. Extending such reasoning to explicit future rollouts at every inference step, however, introduces substantial computational overhead. We propose IG-VLA, a VLA reasoning framework that enables models to imagine the future and internalize the gist. Our Latent Spatiotemporal Reasoning learns to imagine task-relevant future scene evolution directly in visual representation space, guiding action prediction without costly pixel-level video generation. To further reduce inference overhead, we introduce Scene Gist Memory, which internalizes reasoning-derived scene-behavior associations into a compact Scene Gist Token, preserving the benefits of future reasoning while bypassing explicit future imagination at inference. Extensive experiments on LIBERO, LIBERO-Plus, and VLABench demonstrate the effectiveness and efficiency of IG-VLA. On the LIBERO-Plus Language suite, both the reasoning and gist policies outperform the strongest baseline by nearly 6% in success rate. The gist policy also achieves up to 6.38x speedup over baselines, reducing inference latency from 1081ms to 169.5ms per action chunk on a single NVIDIA A6000 GPU. These results demonstrate that future spatiotemporal reasoning can be effectively internalized for efficient VLA deployment.

In this paper, we study how to achieve one-step action generation in Robotic Foundation Models (RFMs), aiming to overcome the high inference latency of multi-step flow matching. MeanFlow provides a promising framework for this goal, yet its direct application leads to performance collapse. We discover that this stems from two distinctive dynamics exhibited in the RFM velocity field: (1) the ``local acceleration" exhibits stability early on, but surges sharply towards the end of the denoising process, and (2) the spread of its magnitudes across samples widens as denoising progresses. To address these issues, we introduce Kinematic MeanFlow (K-MF), a novel one-step action policy tailored for RFMs. Specifically, grounded in a kinematic identity, K-MF decouples the time derivative term in the MeanFlow formulation into two sub-interval terms separated by an intermediate point. This decoupled formulation enables the two terms to capture early-stage and late-stage denoising dynamics, respectively, while mitigating the error amplification across the process. As a result, our K-MF empowers RFMs to achieve one-step action generation in both training from scratch and fine-tuning paradigms across diverse tasks, while outperforming multi-step flow matching in most settings. In terms of inference efficiency, K-MF reduces action-head latency of GR00T-N1.6 by 67.5%~74.4% across L40 and Jetson Orin in eager and compiled modes, yielding end-to-end latency reductions of 30.3%~54.9%. Code will be available at https://github.com/IntelChina-AI/K-MF.

Vision-language-action (VLA) models face a timing gap between low-rate inference and high-rate robot execution. We characterize this gap through end-to-end latency measurements of model inference and the robot execution chain. Repeated Flow Matching denoising contributes substantially to inference cost, while robot-side delays mainly arise from perception acquisition, communication scheduling, and physical response. Analysis of the velocity field shows relatively stable magnitude and direction in early integration, followed by stronger directional correction near the terminal steps. Based on this stage heterogeneity, we propose two-stage non-uniform denoising, reducing the number of steps from 10 to 2 and model-inference time from 61.557 ms to 21.956 ms. We also develop a distributed real-time VLA framework with independent inference, action-publication, and robot-control rates, modular observation acquisition, and action-provenance logging. Using π0.5 as the baseline, we evaluate six real-time execution methods on a long-horizon physical garment-folding task. Legato performs best overall among training-based methods, while Temporal Smoothing leads among training-free methods; both perform strongly in task success, completion time, action continuity, and acceleration smoothness. Combining two-step denoising with representative execution methods substantially reduces inference cost with a small reduction in task performance. These results motivate joint optimization of model-inference efficiency and robot-system timing.

Efficient action generation in vision-language-action (VLA) models requires capturing both coarse action structure and fine-grained details. Discrete action tokens provide compact structural representations but sacrifice precision, while continuous action tokens offer high precision but often require multiple denoising steps. We introduce Discrete Forcing, a flow-matching framework that combines these representations through an explicit coarse-to-fine generation process. It first predicts discrete action tokens to establish a coarse action structure, then uses them to guide continuous action refinement. The discrete and continuous components share a common diffusion transformer backbone with specialized branches, maintaining a parameter count comparable to a conventional single-branch model while requiring only one forward pass per branch. Extensive evaluations across multiple benchmarks demonstrate improved performance and faster inference over a parameter-matched continuous action expert, with consistent performance gains as model capacity increases. Real-world experiments further demonstrate improvements on high-precision and dynamic manipulation tasks.

Visual Language Action (VLA) models offer unprecedented generalization for autonomous robots; however, their real-world deployment is frequently bottlenecked by unreliable execution and the prohibitive computational cost of fine-tuning for specific robot embodiments and tasks. To bridge this gap, we propose WayFinder, an end-to-end, closed-loop hierarchical VLA framework that circumvents the need for fine-tuning by decoupling high-level task reasoning from low-level kinematic control. WayFinder utilizes a zero-shot, offboard Multimodal Large Language Model (MLLM) policy to process linguistic context and state maps for strategic waypoint generation. Asynchronously, a lightweight, onboard policy executes real-time kinematic control at high frequency based on continuous sensor feedback. We evaluate WayFinder in Microsoft AirSim, testing on four environments of varying complexity and three MLLM scales to balance prediction efficacy with computational efficiency. Our results demonstrate that WayFinder achieves superior navigation reliability compared to baseline low-level policies. By querying the high-level MLLM only during navigation failures, WayFinder eliminates the need for fine-tuning, minimizes expensive inferences, and significantly increases navigation success rates by up to 27.45%.

Diffusion and flow-matching Vision-Language-Action (VLA) policies generate action chunks through iterative denoising, incurring substantial inference latency that severely limits real-time robotic control. Existing acceleration methods treat an action chunk as a monolithic computational unit, ignoring a crucial physical reality of receding-horizon control: actions are generated jointly but consumed sequentially, resulting in inherently heterogeneous execution urgencies. We exploit this asymmetry to introduce Urgency-Aware Denoising (UAD), a novel inference-time framework that allocates denoising computation according to when each action is physically needed. UAD releases time-critical urgent actions after fewer denoising steps while overlapping the continued background refinement of tail actions with physical execution. However, heterogeneous denoising introduces two key challenges: early-release errors in urgent actions and trajectory inconsistency in tail actions. UAD elegantly resolves both through two core mechanisms: Trajectory Reconciliation, which reconstructs unified internal state evolution to restore joint denoising coherence without additional model evaluations, and Ghost Action Correction, which leverages non-executed ghost continuations to dynamically compensate for early-release errors across remaining executable actions. Extensive evaluations across multiple VLA architectures, simulation benchmarks, and real-world manipulation tasks demonstrate that UAD achieves up to a 1.89x speedup in average action availability latency while maintaining comparable success rates to vanilla inference with optimal denoising budget, offering a more favorable success-latency trade-off than state-of-the-art VLA acceleration baselines.

Simulated manipulation benchmarks are the standard tool for evaluating vision-language-action (VLA) policies and the acceleration methods that reduce their inference latency for on-robot deployment. On these benchmarks, we observe that some training-free acceleration methods, which approximate the baseline policy's computation, achieve higher measured success rates than the baseline itself. Success rates alone cannot establish whether such gains come from better task execution or from evaluation flaws. We therefore investigate two kinds of benchmark flaws behind these gains: bugs, where the implementation does not match the intended task or evaluation protocol, and design limitations, where success criteria and simulation settings do not fully capture how acceleration affects task execution. Starting from tasks with anomalous gains, we localize root causes by plotting object trajectories against checker acceptance regions, and classify the resulting bugs into task consistency, initialization, and reproducibility. Extending this audit to seven benchmarks, including RoboTwin, LIBERO-Plus, and VLABench, we identify 22 bugs of these types and 4 design limitations. For the latter, we revise permissive success checkers, correct unrealistic object masses, and add a motion-aware score that favors smoother actions. Experiments show that bug fixes can reverse method rankings, moving the baseline from last to first on one task. Addressing design limitations can likewise remove anomalous gains: on another task, the baseline moves from 21 percentage points behind an accelerated method to 5 points ahead. Gains attributed to acceleration can therefore be artifacts of the benchmark rather than better task execution. We release our bug fixes and revised benchmark settings to support trustworthy evaluation of VLA acceleration.

Existing VLA pruning strategies primarily select individual visual tokens according to task-level semantic relevance, while overlooking the spatial information required for robotic manipulation. To examine this limitation, we construct a simple Stride baseline that uniformly samples tokens along the flattened one-dimensional visual sequence, representing a purely geometric pruning strategy. Surprisingly, Stride outperforms semantic pruning and random pruning at certain pruning ratios, but collapses when the token budget is only slightly reduced. We characterize this phenomenon through the spatial coverage radius, defined as the largest spatial blind spot induced by the retained token set after pruning. Our analysis reveals a strong correlation between the spatial structure of retained tokens and task success, suggesting that reliable VLA pruning requires preserving not only task-relevant tokens but also the spatial scaffold of the scene. Motivated by this diagnosis, we propose GeoScaffold, a training-free visual token pruning method that partitions each image into spatial regions, allocates inter-region token budgets using task-relevance weights, and selects intra-region scaffold tokens via farthest point sampling to reduce the local coverage radius. On pi 0.5 and LIBERO, GeoScaffold retains only 20% of visual tokens while preserving a 93.2% average success rate, and achieves a 1.78 times prefill speedup over the unpruned baseline.

Generalist robot policies such as vision-language-action models (VLAs) have achieved remarkable generalization, but their inference delays can conflict with the demands of real-time control. Asynchronous execution avoids pauses between action chunks by predicting the next sequence of actions while the robot carries out the previous one. In this paper, we study whether asynchronous execution produces the same action distribution as the original VLA. We find that, for non-Markovian demonstrations, asynchronous execution can produce a fundamentally different action distribution, which can limit the policy's reactivity. In our method, we seek to restore this reactivity by aligning the asynchronously produced action distribution with that of the original VLA through two complementary mechanisms. First, Recursive Flow-Field Distillation trains the asynchronous policy using the VLA's action-generation flow. We characterize the learned distribution theoretically and show experimentally that our asynchronous policy can generate nearly the full range of actions the original VLA would produce, while existing asynchronous methods recover only a fraction of that range. Second, Propose-Resolve prepares multiple action sequences asynchronously and uses the latest observation to select among them based on a lightweight approximation of their likelihood under the VLA's action distribution. Our resulting method matches the original VLA's success on LIBERO and retains about 80% of its success on RoboMimic, about 30 percentage points more than existing asynchronous methods.

World-Action Models (WAMs) improve robotic manipulation by conditioning action generation on predicted future observations, but future prediction adds further inference overhead to already expensive iterative action generation. Action chunking can amortize this cost over multiple actions, yet performance degrades over long execution horizons because later actions remain conditioned on stale observations. We introduce STAIRCASE POLICY, a streaming inference and training framework that turns a flow-matching VLA into a JEPA-style WAM and partitions a large action chunk into sub-chunks at staggered denoising stages. Near-term actions are executed as soon as they become available, while later actions continue to be refined. At each sub-chunk boundary, the future latent is re-predicted from the latest observation and used to update all unexecuted actions, enabling long-horizon execution without repeated full policy inference. The resulting future-prediction error can further serve as a signal for adaptive chunking. S-WAM achieves 97.7% on LIBERO and 87.9% on LIBERO-Plus, and improves performance across multiple policy backbones and real-robot tasks. It reaches 292.7 executed actions per second, $3.62\times$ the throughput of conventional execution at comparable accuracy, while reducing time-to-first-action from 123.6 to 73.3 ms. With additional inference optimizations, throughput further increases to 642.9 actions per second.

Vision-language navigation (VLN) models perform well but target compute-rich platforms, limiting deployment on memory- and power-constrained robotic edge devices. Compression alone does not establish whether a VLN model fits the memory, latency, and energy budgets of an edge platform while preserving navigation behavior. We introduce EdgeVLN, a runtime-aware, deployment-ready quantized VLN model that closes this gap. EdgeVLN combines a quantized StreamVLN model with Latent Trajectory Termination Extractor (LATTE), a lightweight causal transformer that improves real-time stopping by predicting a Stop Action verifier rank. Both execute through our llama.cpp VLN driver, which reconstructs streaming context and prunes memory tokens on-board. We characterize a pretrained StreamVLN backbone across weight quantization from 8 to 2 bits and multiple inference runtimes to identify a feasible operating point. LATTE reuses backbone hidden states within the budget freed by quantization, requiring neither a second vision encoder nor an additional backbone forward pass. We evaluate six backbone precisions and seven candidate stop heads on BF16 and IQ4 NL across all 1,839 R2R VLN-CE val-unseen episodes. We measure success rate (SR) in simulation and latency, energy, and resident memory on an NVIDIA Jetson Orin NX 16 GB. LATTE achieves our highest SR, 58.02 percent on the deployed 4-bit model, exceeding the BF16 baseline with only 0.013 s additional latency per navigation step. Four-bit formats achieve nearly identical SR, but step energy varies 36.8 times by execution path. Only IQ4 NL under our VLN driver fits the board, using 11.35 GB resident memory while running 20.8 times faster and using 13.3 times less energy than storage-streamed BF16. INT2 collapses. Runtime selection, memory-token pruning, and quantization are essential for efficient edge deployment.

World action models (WAMs) predict the future alongside actions during training. Due to the heavy computation cost of video denoising, whether the future must still be generated during inference is disputed: Explicit WAMs denoise it into clean frames along with every action chunk, whereas Latent WAMs discard it entirely for acceleration. We find that latent WAMs, despite matching explicit ones on in-distribution tasks, fail to retain the generalization benefits that originally motivated WAMs. To demonstrate this, we evaluate generalization along three axes: environmental perturbation, data efficiency, and task generalization. Controlled comparisons with a matched backbone, training data, and budget reveal consistent degradation across all three axes when the action expert no longer conditions on future representations. Further analysis shows that the gap arises almost entirely from the first denoising step: the benefit comes from preparing the future, not generating it. We therefore propose Simple-WAM, which simplifies future modeling into a single forward pass of fully noised video tokens and adapts the training-time noise schedule to this inference behavior. Across simulation and real-world tasks, Simple-WAM achieves the best of both worlds, leading explicit WAMs in generalization performance with efficiency comparable to Latent WAMs. Project Page: https://zrporz.github.io/Simple-WAM-Web/

Modern robot policies predict a chunk of future actions from a single observation, execute only a prefix, and discard the rest before replanning. Choosing the length of this prefix, the execution horizon, poses a trade-off between reactivity and efficiency. A short horizon keeps the policy reactive to the environment, but requires frequent policy calls. Recent test-time methods adaptively select the horizon for each chunk, but they either read model internals, where the signal must be chosen for each architecture, or draw extra samples, which adds cost. We propose Action Upcycling, a training-free algorithm that reuses actions the policy would otherwise discard, without accessing model internals or drawing extra samples. We find that discarded actions stay close to their replanned versions as long as the action velocity remains smooth. Action Upcycling therefore extends the execution horizon up to the point where the velocity begins to fluctuate. Extensive experiments on simulated and real-world manipulation tasks show that Action Upcycling reduces policy calls by 1.2-1.7x with no loss in success rate, across multiple Vision-Language-Action Models (VLAs) and even a World Action Model (WAM). It applies to any chunked policy at negligible cost and is orthogonal to other policy acceleration methods such as few-step sampling and streaming action decoding, opening a new axis for policy acceleration.

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 $π_{0.5}$ 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.

Recent advances in Vision-Language-Action (VLA) models point toward general-purpose robotic intelligence by unifying perception, instruction, and control. Despite impressive progress, existing VLA models often adapt poorly due to \emph{tri-modal misalignment} among vision, language, and action, which weakens action grounding and hurts generalization and fine-tuning efficiency. In this work, we present Alignment-Guided Flow Transformer (AGFT), a novel framework that explicitly enforces tri-modal alignment through a dedicated alignment loss, bridging the representational gap across modalities and enhancing task adaptation. While prior research has predominantly emphasized bi-modal vision--language alignment, we systematically formalize and study tri-modal alignment in VLA models, and provide both ablations and analysis to isolate its role in improving adaptation and robustness. To further accelerate deployment, we adopt a flow-matching objective, enabling substantially fewer inference steps than diffusion-based policies while maintaining accuracy. Theoretically, we establish a quantitative connection between the tri-modal alignment gap and the optimization tightness of flow matching; empirically, experiments on the extensive benchmark show that AGFT achieves superior success rates and lower inference latency compared to SOTA baselines, underscoring tri-modal alignment as a key ingredient for scaling robust VLA manipulation.

Vision-Language-Action (VLA) models enable generalizable robotic control but remain computationally expensive. Token caching provides a training-free, plug-and-play acceleration alternative. However, existing VLA caching does not fully exploit a key inductive bias of VLA models: text-vision synergy, wherein textual semantics guide the precise visual grounding of task-relevant regions. In particular, existing designs insufficiently account for head-wise reliability in attention aggregation and layer-wise stability in cache reuse. To address this, we propose Text-Vision Synergistic Token Caching (TVCache), a training-free framework for efficient VLA inference. TVCache filters attention heads based on text-vision information focus to improve task-relevant and physically consistent visual grounding. Concurrently, we introduce a reuse-layer selection mechanism guided by text-vision entropy differences to avoid caching unstable representations and improve cache resource allocation. Extensive experiments across four representative VLA models, two simulation benchmarks, and real-world robotic tasks demonstrate the effectiveness and generality of TVCache. At matched token-retention ratios, TVCache consistently improves task success over existing VLA caching with comparable computational cost. On OpenVLA-OFT, it improves average success by up to 14.5 percentage points over VLA-Cache at 12.5% retention while reducing FLOPs by 2.45x relative to full-token inference.

Flow-based vision-language-action (VLA) models are highly effective for generalist robot manipulation, yet their reliance on computationally expensive VLM encoding and multi-step iterative action generation imposes a significant latency bottleneck. The resulting inference latency makes it difficult for robots to respond quickly, especially in dynamic environments. We address this limitation with RAVEL (Rolling Asynchronous VLA Enabling Low-Latency Control), an asynchronous inference framework that addresses the computational bottlenecks of both the VLM backbone and the action expert. To reduce the delay from multi-step action denoising, RAVEL allows near-term actions to be executed after a single denoising step by carrying partially denoised future actions forward in a rolling buffer. To avoid blocking on slow VLM encoding, RAVEL decouples VLM encoding from rolling action generation, allowing the action expert to operate continuously using the latest available VLM context, while a lightweight Fast Observation Pathway (FOP) directly conditions the action expert on current observations. Across simulated and real-world manipulation tasks, RAVEL consistently achieves substantially lower response latency while maintaining the task capability of the underlying VLA, enabling high-frequency and responsive closed-loop control.

Vision-Language-Action (VLA) models integrate pretrained Vision-Language Models (VLMs) with action heads for robot control. Common action heads have distinct limitations: point regression provides only a point estimate of the action distribution, while standard flow-matching samplers require costly iterative sampling. To address these limitations, we unify regression and flow matching under a shared objective and extend it to derive a quantile objective. This quantile objective guides the design of our Quantile Head, which predicts a median and positive gaps to form ordered marginal action quantiles in one forward pass. These quantiles support multiple sampling strategies without retraining and are jointly supervised to train the default median policy. Our local analysis of this joint supervision shows that, with calibrated nearby quantiles, fixed gaps, and matched correction speed, direct median updates have lower variance than under median-only supervision. Experiments show that this jointly supervised median policy achieves the highest average success rates among the compared methods on LIBERO, LIBERO-Plus, LIBERO-Pro, and two real-robot tasks, together with the shortest mean episode time among matched LIBERO baselines; code is available at https://github.com/xwangrs/Quantile-Head-for-VLA.