AI & ML
Fast Inference-Time Policy Refinement via Action-Affine World Models
arXiv cs.RO0 views
arXiv:2611.15011v3 Announce Type: replace Abstract: Imitation policies can fail through compounding errors, while existing refinement methods often require additional on-policy learning or expensive online planning. Our key insight is to model the response to small corrections around a policy proposal in an affine form that enables fast optimization. We refine a frozen imitation policy using an \textit{action-affine world model} (A\textsuperscript{2}WM), trained offline from perturbed policy rollouts. Conditioned on history and the proposal, the model directly predicts a nominal trajectory and a linear action response over the full horizon. With quadratic costs and linear constraints, refinement becomes a cheap quadratic program (QP), requiring only one model evaluation and one QP solve per control step. To the best of our knowledge, we are the first to combine an action-affine world model with QP-based inference-time refinement of frozen imitation policies. Across four manipulation tasks, our framework achieves the highest success rate among evaluated baselines, adding only 4.2 ms of mean refinement latency per control step, excluding base-policy sampling.
Source:https://arxiv.org/abs/2611.15011PreviousMOSU2: Long-range Navigation System with Multi-modal Scene UnderstandingNextWhat Can a VLA Learn from the Agent That Calls It? Task-Level Experience Distillation for Embodied Navigation
Related posts
Darwin-Prediction-Weather: Korean AI Model Beats ECMWF, NOAA, and 6 Other Forecast Models Across 28 Benchmark Categories
0dev.to OpenSource

Your use case is not in the benchmarks
0dev.to Programming
Gemini Notebook transforma materiais de estudo em sites interativos
0dev.to Programming

How to Identify Missing Information in Construction Drawings
0dev.to JavaScript
Agenvoy@v1.3.17
0dev.to Go

Orchestrate DuckDB analytics pipelines with Amazon MWAA
0AWS Big Data Blog