Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
CorAl: Introspection for robust radar and lidar perception in diverse environments using differential entropy
Örebro University, School of Science and Technology. (MRO Lab of the AASS Research Centre)ORCID iD: 0000-0003-2504-2488
Örebro University, School of Science and Technology. (MRO Lab of the AASS Research Centre)
Örebro University, School of Science and Technology. (MRO Lab of the AASS Research Centre)ORCID iD: 0000-0001-8658-2985
Örebro University, School of Science and Technology. (MRO Lab of the AASS Research Centre)ORCID iD: 0000-0003-0217-9326
Show others and affiliations
2022 (English)In: Robotics and Autonomous Systems, ISSN 0921-8890, E-ISSN 1872-793X, Vol. 155, article id 104136Article in journal (Refereed) Published
Abstract [en]

Robust perception is an essential component to enable long-term operation of mobile robots. It depends on failure resilience through reliable sensor data and pre-processing, as well as failure awareness through introspection, for example the ability to self-assess localization performance. This paper presents CorAl: a principled, intuitive, and generalizable method to measure the quality of alignment between pairs of point clouds, which learns to detect alignment errors in a self-supervised manner. CorAl compares the differential entropy in the point clouds separately with the entropy in their union to account for entropy inherent to the scene. By making use of dual entropy measurements, we obtain a quality metric that is highly sensitive to small alignment errors and still generalizes well to unseen environments. In this work, we extend our previous work on lidar-only CorAl to radar data by proposing a two-step filtering technique that produces high-quality point clouds from noisy radar scans. Thus, we target robust perception in two ways: by introducing a method that introspectively assesses alignment quality, and by applying it to an inherently robust sensor modality. We show that our filtering technique combined with CorAl can be applied to the problem of alignment classification, and that it detects small alignment errors in urban settings with up to 98% accuracy, and with up to 96% if trained only in a different environment. Our lidar and radar experiments demonstrate that CorAl outperforms previous methods both on the ETH lidar benchmark, which includes several indoor and outdoor environments, and the large-scale Oxford and MulRan radar data sets for urban traffic scenarios. The results also demonstrate that CorAl generalizes very well across substantially different environments without the need of retraining.

Place, publisher, year, edition, pages
Elsevier, 2022. Vol. 155, article id 104136
Keywords [en]
Radar, Introspection, Localization
National Category
Computer Vision and Robotics (Autonomous Systems)
Identifiers
URN: urn:nbn:se:oru:diva-100756DOI: 10.1016/j.robot.2022.104136ISI: 000833416900001Scopus ID: 2-s2.0-85132693467OAI: oai:DiVA.org:oru-100756DiVA, id: diva2:1689786
Funder
Knowledge FoundationEuropean Commission, 101017274Vinnova, 2019-05878Available from: 2022-08-24 Created: 2022-08-24 Last updated: 2024-01-02Bibliographically approved
In thesis
1. Robust large-scale mapping and localization: Combining robust sensing and introspection
Open this publication in new window or tab >>Robust large-scale mapping and localization: Combining robust sensing and introspection
2023 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

The presence of autonomous systems is rapidly increasing in society and industry. To achieve successful, efficient, and safe deployment of autonomous systems, they must be navigated by means of highly robust localization systems. Additionally, these systems need to localize accurately and efficiently in realtime under adverse environmental conditions, and within considerably diverse and new previously unseen environments.

This thesis focuses on investigating methods to achieve robust large-scale localization and mapping, incorporating robustness at multiple stages. Specifically, the research explores methods with sensory robustness, utilizing radar, which exhibits tolerance to harsh weather, dust, and variations in lighting conditions. Furthermore, the thesis presents methods with algorithmic robustness, which prevent failures by incorporating introspective awareness of localization quality. This thesis aims to answer the following research questions:

How can radar data be efficiently filtered and represented for robust radar odometry? How can accurate and robust odometry be achieved with radar? How can localization quality be assessed and leveraged for robust detection of localization failures? How can self-awareness of localization quality be utilized to enhance the robustness of a localization system?

While addressing these research questions, this thesis makes the following contributions to large-scale localization and mapping: A method for robust and efficient radar processing and state-of-the-art odometry estimation, and a method for self-assessment of localization quality and failure detection in lidar and radar localization. Self-assessment of localization quality is integrated into robust systems for large-scale Simultaneous Localization And Mapping, and rapid global localization in prior maps. These systems leverage self-assessment of localization quality to improve performance and prevent failures in loop closure and global localization, and consequently achieve safe robot localization.

The methods presented in this thesis were evaluated through comparative assessments of public benchmarks and real-world data collected from various industrial scenarios. These evaluations serve to validate the effectiveness and reliability of the proposed approaches. As a result, this research represents a significant advancement toward achieving highly robust localization capabilities with broad applicability.

Place, publisher, year, edition, pages
Örebro: Örebro University, 2023. p. 72
Series
Örebro Studies in Technology, ISSN 1650-8580 ; 100
Keywords
SLAM, Localization, Robustness, Radar
National Category
Computer Sciences
Identifiers
urn:nbn:se:oru:diva-107548 (URN)9789175295244 (ISBN)
Public defence
2023-10-31, Örebro universitet, Långhuset, Hörsal L2, Fakultetsgatan 1, Örebro, 13:00 (English)
Opponent
Supervisors
Available from: 2023-08-15 Created: 2023-08-15 Last updated: 2024-01-19Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textScopus

Search in DiVA

By author/editor
Adolfsson, DanielCastellano-Quero, ManuelMagnusson, MartinLilienthal, Achim J.Andreasson, Henrik
By organisation
School of Science and Technology
In the same journal
Robotics and Autonomous Systems
Computer Vision and Robotics (Autonomous Systems)

Search outside of DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 203 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf