Autoregressive Policies Achieve Real-Time Execution in Vision-Language-Action Models
A new preprint from arXiv demonstrates that autoregressive Vision-Language-Action (VLA) models can achieve real-time execution by adjusting tokenization horizons and applying constrained decoding. Prior work on real-time robot control had focused almost exclusively on diffusion-based policies, leaving autoregressive approaches — which are inherently slower in synchronous inference — underexplored for this use case. The findings matter because autoregressive policies offer advantages in convergence speed and instruction-following generalizability, and this work suggests they need not be sacrificed for real-time deployment.
A preprint submitted to arXiv on June 11, 2026 presents a method enabling autoregressive policies to meet strict latency requirements for real-time robotic execution, a capability previously demonstrated mainly for diffusion-based policies. The authors argue that real-time execution is actually more critical for autoregressive models than for diffusion models, since autoregressive rollout is inherently slower under synchronous inference. Their approach combines adjustments to the tokenization horizon with constrained decoding to guarantee latency bounds, which in turn enables multi-trajectory decoding to optimize task performance. Tested across both simulated and real-world environments, the autoregressive policy consistently outperformed an equivalent flow-matching (diffusion-style) policy counterpart and achieved faster task completion compared to synchronous inference baselines. The authors conclude that autoregressive policies retain competitive standing as a policy type for real-time robotics, bolstered by their known strengths in faster convergence and better generalization to novel instructions. The work addresses a meaningful gap in the literature on large-scale VLA model deployment.
What's missing
As a preprint, this work has not yet undergone peer review. The paper does not detail how the constrained decoding approach performs under varying network or compute constraints in deployment. Comparisons are limited to a single flow-matching counterpart, leaving open how the method fares against a broader range of diffusion policy variants.
What different sources said
- arXiv cs.AICenter
Real-Time Execution with Autoregressive Policies
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.