Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Making the Flow Glow - Robot Perception under Severe Lighting Conditions using Normalizing Flow Gradients

Kristoffersson Lind, Simon LU ; Triebel, Rudolph and Krueger, Volker LU orcid (2024) IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2024
Abstract
Modern robotic perception is highly dependent on neural networks.
It is well known that neural network-based perception can be unreliable in real-world deployment,
especially in difficult imaging conditions.
Out-of-distribution detection is commonly proposed as a solution for ensuring reliability in real-world deployment.
Previous work has shown that normalizing flow models can be used for out-of-distribution detection to
improve reliability of robotic perception tasks.
Specifically, camera parameters can be optimized with respect to the likelihood output from a normalizing flow,
which allows a perception system to adapt to difficult vision scenarios.
With this work we propose to use the absolute gradient... (More)
Modern robotic perception is highly dependent on neural networks.
It is well known that neural network-based perception can be unreliable in real-world deployment,
especially in difficult imaging conditions.
Out-of-distribution detection is commonly proposed as a solution for ensuring reliability in real-world deployment.
Previous work has shown that normalizing flow models can be used for out-of-distribution detection to
improve reliability of robotic perception tasks.
Specifically, camera parameters can be optimized with respect to the likelihood output from a normalizing flow,
which allows a perception system to adapt to difficult vision scenarios.
With this work we propose to use the absolute gradient values from a normalizing flow,
which allows the perception system to optimize local regions rather than the whole image.
By setting up a table top picking experiment with exceptionally difficult lighting conditions,
we show that our method achieves a 60% higher success rate for an object detection task
compared to previous methods. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
host publication
2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
publisher
IEEE - Institute of Electrical and Electronics Engineers Inc.
conference name
IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2024
conference location
Abu Dabi, United Arab Emirates
conference dates
2024-10-14 - 2024-10-18
external identifiers
  • scopus:85216480088
ISBN
979-8-3503-7770-5
DOI
10.1109/IROS58592.2024.10801601
language
English
LU publication?
yes
id
f2096f1c-9153-40b5-aec4-503edfa2804a
date added to LUP
2024-11-15 12:02:14
date last changed
2025-04-04 13:53:28
@inproceedings{f2096f1c-9153-40b5-aec4-503edfa2804a,
  abstract     = {{Modern robotic perception is highly dependent on neural networks.<br/>It is well known that neural network-based perception can be unreliable in real-world deployment,<br/>especially in difficult imaging conditions.<br/>Out-of-distribution detection is commonly proposed as a solution for ensuring reliability in real-world deployment.<br/>Previous work has shown that normalizing flow models can be used for out-of-distribution detection to<br/>improve reliability of robotic perception tasks.<br/>Specifically, camera parameters can be optimized with respect to the likelihood output from a normalizing flow,<br/>which allows a perception system to adapt to difficult vision scenarios.<br/>With this work we propose to use the absolute gradient values from a normalizing flow,<br/>which allows the perception system to optimize local regions rather than the whole image.<br/>By setting up a table top picking experiment with exceptionally difficult lighting conditions,<br/>we show that our method achieves a 60% higher success rate for an object detection task<br/>compared to previous methods.}},
  author       = {{Kristoffersson Lind, Simon and Triebel, Rudolph and Krueger, Volker}},
  booktitle    = {{2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}},
  isbn         = {{979-8-3503-7770-5}},
  language     = {{eng}},
  month        = {{12}},
  publisher    = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
  title        = {{Making the Flow Glow - Robot Perception under Severe Lighting Conditions using Normalizing Flow Gradients}},
  url          = {{http://dx.doi.org/10.1109/IROS58592.2024.10801601}},
  doi          = {{10.1109/IROS58592.2024.10801601}},
  year         = {{2024}},
}