Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Normalizing flows are capable models for bi-manual visuomotor policy

Li, Jialong LU orcid ; Kristoffersson Lind, Simon LU ; Xie, Wenrui LU ; Stenmark, Maj LU orcid and Krueger, Volker LU orcid (2026)
Abstract
The field of general-purpose robotics has recently embraced powerful probabilistic diffusion-based models to learn the complex embodiment behaviours. However, existing models often come with significant trade-offs, namely high computational costs for inference and a fundamental inability to quantify output uncertainty. We introduce Normalizing Flows Policy (NF-P), a conditional normalizing flow-based visuomotor policy for bi-manual manipulation. NF-P learns a conditional density over action sequences and enables single-pass generative sampling with tractable likelihood computation. Using this property, we propose two inference-time optimization strategies: Stochastic Batch Selection, which selects the highest-likelihood trajectory among... (More)
The field of general-purpose robotics has recently embraced powerful probabilistic diffusion-based models to learn the complex embodiment behaviours. However, existing models often come with significant trade-offs, namely high computational costs for inference and a fundamental inability to quantify output uncertainty. We introduce Normalizing Flows Policy (NF-P), a conditional normalizing flow-based visuomotor policy for bi-manual manipulation. NF-P learns a conditional density over action sequences and enables single-pass generative sampling with tractable likelihood computation. Using this property, we propose two inference-time optimization strategies: Stochastic Batch Selection, which selects the highest-likelihood trajectory among sampled candidates, and Gradient Refinement, which directly ascends the log-likelihood to improve action quality. In both simulation and real robot experiments, NF-P achieves promising success rates compared to the baseline. In addition to improved task performance, NF-P demonstrates faster training and lower inference latency. These results establish normalizing flows as a competitive and computationally efficient visuomotor policy, particularly for real-time, uncertainty-aware robotic control. (Less)
Please use this url to cite or link to this publication:
author
; ; ; and
organization
publishing date
type
Working paper/Preprint
publication status
submitted
subject
pages
8 pages
language
English
LU publication?
yes
id
9e060c64-620b-49ba-b1c8-495e624f875c
alternative location
https://arxiv.org/abs/2509.21073
date added to LUP
2026-04-10 14:16:04
date last changed
2026-05-19 12:35:57
@misc{9e060c64-620b-49ba-b1c8-495e624f875c,
  abstract     = {{The field of general-purpose robotics has recently embraced powerful probabilistic diffusion-based models to learn the complex embodiment behaviours. However, existing models often come with significant trade-offs, namely high computational costs for inference and a fundamental inability to quantify output uncertainty. We introduce Normalizing Flows Policy (NF-P), a conditional normalizing flow-based visuomotor policy for bi-manual manipulation. NF-P learns a conditional density over action sequences and enables single-pass generative sampling with tractable likelihood computation. Using this property, we propose two inference-time optimization strategies: Stochastic Batch Selection, which selects the highest-likelihood trajectory among sampled candidates, and Gradient Refinement, which directly ascends the log-likelihood to improve action quality. In both simulation and real robot experiments, NF-P achieves promising success rates compared to the baseline. In addition to improved task performance, NF-P demonstrates faster training and lower inference latency. These results establish normalizing flows as a competitive and computationally efficient visuomotor policy, particularly for real-time, uncertainty-aware robotic control.}},
  author       = {{Li, Jialong and Kristoffersson Lind, Simon and Xie, Wenrui and Stenmark, Maj and Krueger, Volker}},
  language     = {{eng}},
  note         = {{Preprint}},
  title        = {{Normalizing flows are capable models for bi-manual visuomotor policy}},
  url          = {{https://arxiv.org/abs/2509.21073}},
  year         = {{2026}},
}