AI & ML
LoRN: Low-Rank Noise Steering for Generative Robot Policies
arXiv cs.RO0 views
arXiv:2611.23628v1 Announce Type: new Abstract: Noise steering improves pretrained generative robot policies by learning to select their initial noise without updating model weights. We find that actions in the evaluated world-action models are more sensitive to video noise than action noise, making video noise an expressive input for steering. However, the noise-steering methods we evaluate, in their published forms, struggle to scale to the high-dimensional noise inputs of large policy classes such as world-action models. We introduce Low-Rank Noise Steering (LoRN), a reinforcement learning method that scales noise steering to world-action models. A small steering actor selects a direction through a fixed low-rank random projection, while our norm-controlled construction keeps the resulting noise near the typical scale of a Gaussian sample. This separates the actor's control dimension from the full noise dimension, enabling steering of both video and action inputs. Across three world-action models and ten model-task pairs, video steering improves average success by 26.2 percentage points over the base policy, compared with 7.1 points for DSRL action steering. For smaller action policies, the same construction achieves comparable final success while generally improving early-training success. Steering actors trained in simulation also transfer to three physical bimanual manipulation tasks across two base policies, without additional real-world training.
Source:https://arxiv.org/abs/2611.23628PreviousHow Uncertain Is My Action? Self-Correcting Visual Planning with Frozen World ModelsNextRadar-Native Control Barrier Functions: Doppler Range Rate as a Measured Barrier Derivative
Related posts
Anthropic soars in Wall Street debut despite AI jitters
0Phys.org Tech
Making graphene work in metal manufacturing
0Phys.org Tech

Job applicants know AI is reading their CV — and are rewriting the files to get past the first hurdle
0TechRadar

Apple TV reveals new series with Andrew Garfield, Jude Law as magicians
09to5Mac

Gemini app for Mac adds computer use to control desktop apps
09to5Google
At $600, the Pixel 10 is this year’s no-brainer Black Friday smartphone deal
09to5Google