Devil in the details: Systematic review of TOR signals in automated driving with a generic classification framework

Reinier J. Jansen; Angelica M. Tinga; Rins de Zwart; Sander T. van der Kint

doi:10.1016/j.trf.2022.10.009

Devil in the details: Systematic review of TOR signals in automated driving with a generic classification framework

Reinier J. Jansen^*
, Angelica M. Tinga
, Rins de Zwart
, Sander T. van der Kint

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

12 Citations (Scopus)

159 Downloads (Pure)

Abstract

Background: Meta studies on factors contributing to take-over performance did not include the design of take-over request (TOR) signals, other than the modality at which TORs are presented. A detailed understanding of the influence of TOR design on take-over performance is therefore lacking.

Objective: To gain an overview of the level of detail with which TOR designs are reported in academic literature, by using and evaluating a novel classification framework. In this framework TORs are classified in terms of modalities, classes, and underlying attributes. Furthermore, the framework involves classification of potentially competing background signals, as well as the setting in which a study is performed.

Method: A systematic review was performed on articles written in English that were published between 2014 and 2021 using Web of Science, as well as articles retrieved from two previous TOR classification studies and three meta studies on take-over performance. Studies were considered for subsequent analysis if they involved a downward transition of the level of automation following a TOR, resulting in a sample of 391 TORs found in 189 studies.

Results: No predominant TOR design was found, and a considerable part of the available design space has not yet been explored. Studies reported less information on TOR designs when examining TOR designs at an increased level of detail. On average, attribute information was reported for half of the TORs per class.

Conclusions: More attention towards a detailed description of TOR implementations is needed and how this can impact experimental findings. The classification framework and the corresponding coding sheet could support systematic reporting and subsequent meta-analysis in future work. This way, a better understanding about the impact of TOR design on take-over performance can be gained, which in turn can support implementation of safe and effective TORs in (automated) vehicles.

Original language	English
Pages (from-to)	274-328
Number of pages	55
Journal	Transportation Research Part F: Traffic Psychology and Behaviour
Volume	91
DOIs	https://doi.org/10.1016/j.trf.2022.10.009
Publication status	Published - Nov 2022
Externally published	Yes

Keywords

Conceptual framework
Design alternatives
Human-machine interface
Implementation details
Systematic literature review
Take-over request

Access to Document

10.1016/j.trf.2022.10.009Licence: CC BY

ArticleFinal published version, 5.31 MBLicence: CC BY

Cite this

@article{d6357da88ea24998a5a6b37a0f041c47,

title = "Devil in the details: Systematic review of TOR signals in automated driving with a generic classification framework",

abstract = "Background: Meta studies on factors contributing to take-over performance did not include the design of take-over request (TOR) signals, other than the modality at which TORs are presented. A detailed understanding of the influence of TOR design on take-over performance is therefore lacking.Objective: To gain an overview of the level of detail with which TOR designs are reported in academic literature, by using and evaluating a novel classification framework. In this framework TORs are classified in terms of modalities, classes, and underlying attributes. Furthermore, the framework involves classification of potentially competing background signals, as well as the setting in which a study is performed.Method: A systematic review was performed on articles written in English that were published between 2014 and 2021 using Web of Science, as well as articles retrieved from two previous TOR classification studies and three meta studies on take-over performance. Studies were considered for subsequent analysis if they involved a downward transition of the level of automation following a TOR, resulting in a sample of 391 TORs found in 189 studies.Results: No predominant TOR design was found, and a considerable part of the available design space has not yet been explored. Studies reported less information on TOR designs when examining TOR designs at an increased level of detail. On average, attribute information was reported for half of the TORs per class.Conclusions: More attention towards a detailed description of TOR implementations is needed and how this can impact experimental findings. The classification framework and the corresponding coding sheet could support systematic reporting and subsequent meta-analysis in future work. This way, a better understanding about the impact of TOR design on take-over performance can be gained, which in turn can support implementation of safe and effective TORs in (automated) vehicles.",

keywords = "Conceptual framework, Design alternatives, Human-machine interface, Implementation details, Systematic literature review, Take-over request",

author = "Jansen, \{Reinier J.\} and Tinga, \{Angelica M.\} and \{de Zwart\}, Rins and \{van der Kint\}, \{Sander T.\}",

note = "Funding Information: The authors declare that they have no conflict of interest. This study was funded by the Dutch Ministry of Infrastructure and Water Management. The funder played no role in study design, collection, analysis, interpretation of data, writing of the report, or in the decision to submit the paper for publication. The funder accepts no responsibility for the contents. Publisher Copyright: {\textcopyright} 2022 The Author(s)",

year = "2022",

month = nov,

doi = "10.1016/j.trf.2022.10.009",

language = "English",

volume = "91",

pages = "274--328",

journal = "Transportation Research Part F: Traffic Psychology and Behaviour",

issn = "1369-8478",

publisher = "Elsevier",

}

TY - JOUR

T1 - Devil in the details

T2 - Systematic review of TOR signals in automated driving with a generic classification framework

AU - Jansen, Reinier J.

AU - Tinga, Angelica M.

AU - de Zwart, Rins

AU - van der Kint, Sander T.

N1 - Funding Information: The authors declare that they have no conflict of interest. This study was funded by the Dutch Ministry of Infrastructure and Water Management. The funder played no role in study design, collection, analysis, interpretation of data, writing of the report, or in the decision to submit the paper for publication. The funder accepts no responsibility for the contents. Publisher Copyright: © 2022 The Author(s)

PY - 2022/11

Y1 - 2022/11

N2 - Background: Meta studies on factors contributing to take-over performance did not include the design of take-over request (TOR) signals, other than the modality at which TORs are presented. A detailed understanding of the influence of TOR design on take-over performance is therefore lacking.Objective: To gain an overview of the level of detail with which TOR designs are reported in academic literature, by using and evaluating a novel classification framework. In this framework TORs are classified in terms of modalities, classes, and underlying attributes. Furthermore, the framework involves classification of potentially competing background signals, as well as the setting in which a study is performed.Method: A systematic review was performed on articles written in English that were published between 2014 and 2021 using Web of Science, as well as articles retrieved from two previous TOR classification studies and three meta studies on take-over performance. Studies were considered for subsequent analysis if they involved a downward transition of the level of automation following a TOR, resulting in a sample of 391 TORs found in 189 studies.Results: No predominant TOR design was found, and a considerable part of the available design space has not yet been explored. Studies reported less information on TOR designs when examining TOR designs at an increased level of detail. On average, attribute information was reported for half of the TORs per class.Conclusions: More attention towards a detailed description of TOR implementations is needed and how this can impact experimental findings. The classification framework and the corresponding coding sheet could support systematic reporting and subsequent meta-analysis in future work. This way, a better understanding about the impact of TOR design on take-over performance can be gained, which in turn can support implementation of safe and effective TORs in (automated) vehicles.

AB - Background: Meta studies on factors contributing to take-over performance did not include the design of take-over request (TOR) signals, other than the modality at which TORs are presented. A detailed understanding of the influence of TOR design on take-over performance is therefore lacking.Objective: To gain an overview of the level of detail with which TOR designs are reported in academic literature, by using and evaluating a novel classification framework. In this framework TORs are classified in terms of modalities, classes, and underlying attributes. Furthermore, the framework involves classification of potentially competing background signals, as well as the setting in which a study is performed.Method: A systematic review was performed on articles written in English that were published between 2014 and 2021 using Web of Science, as well as articles retrieved from two previous TOR classification studies and three meta studies on take-over performance. Studies were considered for subsequent analysis if they involved a downward transition of the level of automation following a TOR, resulting in a sample of 391 TORs found in 189 studies.Results: No predominant TOR design was found, and a considerable part of the available design space has not yet been explored. Studies reported less information on TOR designs when examining TOR designs at an increased level of detail. On average, attribute information was reported for half of the TORs per class.Conclusions: More attention towards a detailed description of TOR implementations is needed and how this can impact experimental findings. The classification framework and the corresponding coding sheet could support systematic reporting and subsequent meta-analysis in future work. This way, a better understanding about the impact of TOR design on take-over performance can be gained, which in turn can support implementation of safe and effective TORs in (automated) vehicles.

KW - Conceptual framework

KW - Design alternatives

KW - Human-machine interface

KW - Implementation details

KW - Systematic literature review

KW - Take-over request

UR - https://www.scopus.com/pages/publications/85140325921

U2 - 10.1016/j.trf.2022.10.009

DO - 10.1016/j.trf.2022.10.009

M3 - Article

SN - 1369-8478

VL - 91

SP - 274

EP - 328

JO - Transportation Research Part F: Traffic Psychology and Behaviour

JF - Transportation Research Part F: Traffic Psychology and Behaviour

ER -

Devil in the details: Systematic review of TOR signals in automated driving with a generic classification framework

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this