Do Researchers Use the Levels of Automation Appropriately? How Automation is Discussed in Research

Abstract

Despite multiple taxonomies and descriptions of automation there is inconsistency in describing automation capabilities, making it difficult to interpret and replicate research. We conducted a systematic literature review to investigate how studies document automation. The Scopus^® database was searched on January 13th, 2023, for vehicle automation studies published in the Proceedings of the Human Factors and Ergonomics Society Conference between 2012 and 2022. Twenty-one studies were identified. Sixteen of these studies described the vehicle automation in the protocol and apparatus and five were missing at least one element. We show inconsistency and insufficiency of these explanations, suggesting researchers and participants might not understand the levels of automation. We offer a guide for improving how researchers describe automation capabilities to improve the interpretability and replicability of studies.

Keywords

Levels of Automation Automation

Introduction

There is no unified taxonomy for describing all variations of motor vehicle automation despite multiple attempts (Gasser & Westhoff, 2012; National Highway Traffic Safety Administration, 2013; SAE International, 2021). The most cited source for describing vehicle automation is the Society of Automotive Engineers (SAE) Levels of Driving Automation (SAE International, 2021). The SAE taxonomy partitions vehicle automation functions into six levels, each with driver responsibilities and automation capabilities (Table 1).

Table 1.

The role of the human and automation system according to SAE Levels of Automation (adapted from J3016).

Automation	Role of the Driver	Role of the Automation
Levels 0-2: The driver performs part or all of the DDT
Level 0	Driver (at all times): Performs the entire DDT	Automation: Does not perform any part of the DDT on a sustained basis (May provide warnings and momentary assistance)
Level 1	Driver (at all times): performs the DDT not performed by the automation system; supervises the system, determines when engagement /disengagement is appropriate; immediately performs the entire DDT when required or desired	Automation: Performs part of the DDT with lateral or longitudinal control; disengages upon request
Level 2	Driver (at all times): performs DDT not performed by the automation; supervises the system, determines when engagement and disengagement is appropriate; Performs all DDT when required or desired	Automation: Performs part of the DDT with lateral and longitudinal control; disengages upon request
Level 3	Driver (ADS not engaged): Determines whether to engage DDT fallback-ready user (ADS engaged): Receptive to requests to intervene and perform DDT; Receptive to DDT performance relevant systems; Determines whether and how to achieve a minimal risk condition	ADS (when in ODD): Performs the entire DDT within its ODD; Requests intervention when leaving ODD and in system failures; Disengages upon user request
Levels 3-5: ADS performs the entire DDT while engaged
Level 4	Passenger/Dispatcher (ADS engaged): May perform DDT after ADS reaches ODD limit; May become a driver	ADS (when in ODD): Performs the entire DDT; May prompt the passenger to resume control at the limits of the ODD
Level 5	Passenger/Dispatcher (ADS engaged): May perform DDT after reaching ODD limit; May become a driver	ADS (while engaged): Performs the entire DDT under driver-manageable on-road conditions

The SAE framework can help communicate automation capabilities. However, drivers and researchers struggle to understand the user’s role and the automation capabilities in terms of the SAE levels (Novakazi et al., 2021). This might undermine the interpretation and replication of research on driver interaction with increasingly automated vehicles and varied capabilities within levels. Although general guidelines help researchers describe their experiments to ensure replication, the complexity of the SAE levels presents challenges. Specifically, they were designed as a reference for engineers and are often difficult for drivers to understand. This difference in understanding can be seen in the results of studies in which driver mental models do not align with the SAE levels (Novakazi et al., 2021). Likewise, researchers miscite the SAE levels when they state nonexistent levels of automation capabilities like “Level 2.5” (Ataya et al., 2021). The SAE levels are framed around three descriptive dimensions, the Dynamic Driving Task (DDT), Object Event Detection and Response (OEDR), and the Operational Design Domain (ODD). DDT is the operational (e.g., steering and acceleration) and tactical (e.g., honking horn, maneuvering, and signaling) functions required to operate a vehicle. The OEDR defines the sensing and control abilities to monitor and respond to the environment, which may include planning maneuvers. The ODD defines automation where and when the automation is meant to operate (e.g., time of day, roadway type, speed, etc.).

The SAE levels specify increasingly capable automation systems, each prescribing different driver roles. Level 0 only provides warnings and momentary assistance using vehicle safety subsystems such as antilock braking and stability control.

Level 1 requires driver-supervised automated steering or braking and performs the remainder of the DDT. Level 2 provides simultaneous longitudinal and lateral control, and the driver performs the remainder of the DDT. Level 3 performs the entire DDT in an ODD but assumes that the driver is receptive to a request to intervene. Level 4 can operate without driver engagement in a defined ODD. Level 5 can perform the entire DDT in any driver-manageable on-road conditions. Importantly, a manufacturer may generally represent the driving automation of their vehicle according to these SAE levels. However, the actual implementation of automation is qualified by the specifics of the Tactical Maneuver Competency of the DDT, the ODD, and the OEDR, within each SAE level. Therefore, driver roles can vary considerably between offerings within an SAE level according to the degree of competency described by each dimension. Specifically, each cell in a threedimensional framework of DDT, OEDR, and ODD define specific variance within each SAE level. On one hand, a framework with fewer discrete levels for each dimension may describe fewer potential systems. On the other hand, a finer scale of each dimension would describe a very large number of potential design variants, each with slightly different driver roles. This implies that the three-dimensional design space, whatever the specificity, will contain regions of variants associated with each SAE level. A key challenge is how to scale the framework for designing driver roles and describing research.

Although this taxonomy seems to unambiguously describe vehicle automation, in practice both researchers and drivers often struggle to appreciate the nuances within and between levels (Novakazi et al., 2021). For example, within Level 2, a vehicle may include features like intelligent speed control and automatic lane changes. Similarly, design affordances within Level 2 may vary across systems, for example, one design may afford a great deal of steering input, while another may afford very little. This paper addresses two research questions: 1) Do the method sections of HFES conference papers precisely describe the automation implemented for the study? and 2) Do these method sections precisely describe the automation to study participants? By addressing these questions, we identify how to precisely document research and experimental conditions.

Methods

Seventy-five articles from the Proceedings of the Human Factors and Ergonomics Society Conference were found using the keywords “driving” OR “vehicle” AND “automation” in the Scopus database. We included studies about vehicle automation in either a driving simulator or automated vehicle, and excluded all other study designs, such as survey studies or literature reviews, reducing this list to 21 articles.

The lead author then screened each article and extracted the explanation of the automation capabilities from the apparatus and protocol. The apparatus describes the automation’s capabilities. Automation capability refers to the control that automation exerts, under what conditions, and if the human must take over when prompted. The protocol instructions are the written description of how the participant was introduced, taught, or trained on how to use the automation.

This includes the participants’ responsibility and the automation capabilities once engaged.

The completeness of the information in the apparatus and protocol sections of the experimental methods were coded. We recorded whether the SAE levels were mentioned, and if the level of automation used in the experiment was described. We also rated how well the description covered information that may support appropriate understanding and trust in the automation: purpose, process, and performance (Lee & See, 2004). These dimensions were defined as: Purpose (P1)—What is the drivers’ role and why the automation was developed? Process (P2)—How it works (e.g., sensors, limits, etc.)? Performance (P3)—What should the user expect from warnings and how are they expected to interact? We rated how completely these dimensions are addressed on a 0 to 3 scale: 0—corresponding to the component’s absence, 1—present but scarce (e.g., states “capable of automated driving” but no details on what it can do), 2—partial details (e.g., States “Level 3” without stating the ODD), 3—enough detail to replicate the conditions.

Results

Table 2 summarizes how 21 papers published in the HFES conference proceedings from 2012–2022 applied the levels of automation and described the vehicle automation. Of these, 14% (3) failed to describe the apparatus, 9.5% (2) did not provide their protocol instructions, 42% cited the SAE levels of automation, and 21.8% (5) explicitly stated the level of automation used.

Table 2.

Explanations of automation capabilities to both the reader (apparatus) and the research participant (protocol). ◊ = Mentions SAE levels standard, ◑ = No participant explanation, ◐ = No automation explanation, ▲ = States level of automation. * = includes an explanation from a prior paper, (Clark &Feng, 2016).

Citation	Level	How automation is described to participants in the protocol	How automation is explained in apparatus	P1	P2	P3
Yang et al 2018 ◊	Level 3 ▲	Participants familiarized themselves with the system in a practice drive in manual mode and the level 3 automated mode.	Level 3 automation.	2	1	1
Clark et al 2017 ◊*	Level 3 ▲	Participants did an introduction session and a practice drive (four transitions between automated and manual) to gain familiarity.	States the simulator is a level 3 automation system; Goes on to describe transitions as “fully automated” to manual	2	1	1
Radlmayr et al 2018 ◊	Level 3 ▲	Participants drove a familiarization drive to adjust to vehicle dynamics, experience the automation, and a take-over.	Level 3 automation which could change lanes and overtake slow traffic	1	2	2
Cohen-Lazry et al 2017 ◊	Level 3 ▲	Participants drove a 7-minute familiarization drive, which included both autonomous and manual driving	Level 3 automation is capable of driving without driver intervention or monitoring until its operating boundaries are met.	1	2	1
Miller & Boyle 2017	Level 1	Participants came in three separate times and there was a practice drive at the beginning of each day.	The lane-keeping assistance system provided full lateral control; participants only had to provide longitudinal control.	1	2	1
Kerschbaum et al 2014	Level 2	Participants were instructed about automation features and took a 10minute test drive to familiarize themselves with automated driving.	The automation was able to perform longitudinal and lateral control and complied with speed limits on the simulated freeway.	1	2	3
Miller et al 2015	Level 3	Participants were told how to engage automation.	States that there is an ADAS.	0	1	0
Shen & Neyens 2014	Levels 1 and 2	Two practice drives before the experiment: one to experience driving manually and one to experience the automated system.	Participants were assigned to one of the two conditions; lane Keep or Adaptive Cruise Control and Lane Keep engaged.	1	1	1
Glaser et al. 2017	Level 3	Participants were informed that the automated driving feature had limitations and required them to monitor and intervene to avoid hazards. The training was conducted to establish what a hazard is and practice identifying and responding to hazards.	Individuals were seated in a partially automated‚ self-driving vehicle capable of controlling its speed, spacing to lead vehicles, and lateral lane position.	2	2	2
Regan et al 2017	Level 3	The researcher summarized each crash avoidance and driver assistance technology including descriptions of each systems’ purpose, operational boundaries, settings, and instructions about use.	Used 5 vehicles with ACC systems that could come to a complete stop and adjust the speed and headway to the lead vehicle. Two Audis and Honda had speed limit recognition and lane departure assistance. The Audis also had active lane-keeping assistance and congestion assist.	1	2	0
Funkhouser & Drews 2016	Level 3	Participants got to do a “warmup scenario” similar to the actual scenario and were allowed to ask questions.	“The car then autonomously drove itself for a designated amount of time”	1	0	0
Schmidt et al 2017	Level 3	Participants drove a 17 min manual driving phase and 8 min automated function phase to get adjusted to the simulator dynamics; alertness and take-over requests were triggered.	Automation controlled the lane-keeping and drove in the longitudinal direction on its own; drivers were always able to take control.	2	1	3
Gold et al 2013	Level 5	Subjects were able to drive manually and learn how to turn the automation on and off in a familiarization drive.	Simulated a “highly automated” which could control longitudinal, lateral, lane changing, and overtaking vehicles.	1	1	0
Blommer et al 2017◐ ◊	Level 2 or 3	Participants were given an overview of the automated driving system’s capabilities. They were allowed to change lanes and adapt to the automated and manual driving modes.	None described.	1	1	0
Blommer et al 2015 ◐ ◊	Level 2 or 3	Participants were given an overview of the automated driving system’s capabilities and the chance to practice automated driving.	None described.	1	1	0
Gold et al 2015 ◐	Level 3	Participants completed practice drives to get familiar with the driving simulation and the automation functions.	None described.	1	0	0
Feldhütter et al 2016 ◑	Level 3	Participants were trained on drivers’ responsibilities, system capabilities, and limits.	Automation took over longitudinal and lateral control and could perform lane changes based on traffic conditions.	2	3	0
Morris et al 2017 ◑	Levels 2 and 4	None described.	The vehicle was programmed to “obey all the rules of the road” in one of the scenarios, and to speed and drift within its lane in another	0	1	0
Neubauer et al 2014 ◑	Levels 2 and 5	None described.	Two levels of automation: Total Automation - steering, braking, and speed; Partial Automation (cruise control) - braking and speed	0	1	0
Biondi et al 2016◊	Level 2	Instructions were given on how to activate, use, and adjust automation	Autopilot and autosteer maintain speed, lane centering, car following with a set distance, and change lanes when asked by the driver.	2	2	0
Miller et al 2016	Levels 2	Participants were provided information on the automated driving system via tablet.	Two automation levels: high - handles more situations without driver intervention; low - requires more driver intervention.	1	1	1

The participant protocol instructions were rarely discussed. The most defined instruction was “familiarization” or practice drives where the participant used the automation before the experiment began. Thirteen studies described that the participant was allowed to explore and use the automation in a familiarization drive before the study began. Four of these study procedures specified that they instructed the participant on the automation’s capabilities and limitations. However, only two described instructing the driver on how to use the automation, and three stated that they gave them “information” about the automation system such as how to turn it on and off.

Discussion

Contents of the Method Sections

The method section is integral to understanding the study and replicating the research and should be thoughtfully written (Coverdale et al., 2006; Kallet, 2004; Kotz & Cals, 2013). It ensures that the experiment is described in enough detail to be replicated by others, allowing readers to judge the validity and generalizability and decide whether to incorporate the findings into practice (Kallet, 2004). When this information is missing fellow researchers cannot reproduce experiments, introduce new variations, or build upon theory. This section also helps others interpret findings and assess validity, such as biasing factors, and allows readers to decide whether to incorporate the findings into practice (Coverdale et al., 2006; Kallet, 2004).

We recommend, when appropriate, researchers write a detailed account of the participants’ introduction to the automation system. When there is limited space, such as with the HFES conference papers, researchers must be concise in describing the automation capabilities. Our guide in Table 3, shows what needs to be discussed for the purpose, process, and performance of the automation to be sufficiently described.

Table 3.

A guide describing automation capabilities and how they are introduced to participants in research papers.

Level	Purpose	Process	Performance
Level 0	State that participants were in full control of the vehicle and must handle all DDT.	It is not applicable to provide details about the process.	It is not applicable to provide details about performance.
Level 1	State that participants were trained on how to use automation features and their intended use (i.e., lane keep helping maintain lane position). Emphasize that they are in control.	Provide details about the automated features. For example, explain that lane keep assist helps keep the vehicle within its lane by preventing any drifting into other lanes.	Describe the expected warnings, if any. For example, if the driver exceeds the lane boundaries, mention that the system may provide warnings and the driver must correct the deviation.
Level 2	State that participants were trained to use automation and that they need to monitor the system.	Provide details about the automation’s lateral and longitudinal control capabilities. Explain that the driver must monitor the system.	Describe the expected warnings and takeover prompts. State if driver can regain control at any time and should intervene if there is a system failure.
Level 3	State that participants were trained on using the automation, emphasizing that it controls all DDT under specific conditions. Emphasize that they need to be prepared to take over.	Provide details about the automation’s capabilities, such as lane changing or overtaking vehicles. Explain that the driver is not required to monitor the system.	Describe the expected warnings and takeover prompts, if any. State if driver can regain control at any time and should intervene if there is a system failure.
Level 4	Clearly state that participants received training emphasizing the automation’s ability to control all DDT, while under specific conditions, without their control or supervision.	Describe the automation’s capabilities, such as handling all DDT within a specified location (e.g., San Francisco). Explain the driver is not required to monitor the system but may be prompted to take over outside the ODD.	Describe the expected warnings and prompts, if any. State if driver can regain control at any time. Explain that if they don’t take over in the event of a failure, the vehicle will achieve a minimal crash risk (MCR).
Level 5	Clearly state that participants received training emphasizing that the automation is capable of driving without any intervention.	Provide details about the automation’s ability to control all driving tasks. Explain that the driver is not required to monitor the system.	Describe the expected warnings, if any. State if driver can regain control at any time. Explain that in the event of a failure, system will achieve MCR.

A more general issue, affecting human subject research is the “psychologist’s fallacy”, a cognitive bias that describes how experimenters mistakenly assume their perception of a situation matches the participants (James, 1890). This is particularly critical in vehicle automation studies because drivers’ mental models are unlikely to be based on SAE levels and so may diverge from those of the SAE (Novakazi et al., 2021; Lee et al., In Press). Mental model quizzes can assess participants’ understanding of the automation and mitigate the psychologist’s fallacy.

The protocol describing the automation should specify the degree of interaction between the participant and the automation. The introduction is important as it indicates how “in the loop” and educated the participant is. A driver’s understanding and trust in automation can change how it is used (Körber et al., 2018; Novakazi et al., 2021). Naturally, researchers want to ensure that the participant understands what is expected of them and will often incorporate a practice drive, where the participant has an opportunity to become familiar with the automation capabilities. However, in studies involving control and experimental groups, participants may or may not be introduced to the automation, given prior hands-on exposure, or told certain details. This further raises the importance of describing any interface and associated design affordances as well as following strict protocols when communicating with subjects. The explanation to participants is critical because it can affect user trust, which influences automation acceptance, monitoring, and reliance (Choi & Ji, 2015; Körber et al., 2018; Lee & See, 2004; Sanders et al., 2019).

This brief review only included articles from the HFES conference proceedings where the five-page limit constrains what researchers can include. Future reviews should broaden the scope and consider journals that provide more space for discussing experimental design. Furthermore, the Scopus^® database does not include all relevant published articles from 2012 to 2022. Future studies should use the SAGE Journals^® database as it contains all HFES papers.

Automation Descriptions in Research

Although SAE levels provided a structure to classify and describe automation, many studies have found that a humancentric description of automation may be more valuable (Jamieson & Skraaning, 2018; Kaber, 2018; Novakazi et al., 2021; J. H. Yang et al., 2017). The current classification of automation levels does not fully account for human behavior or accommodate the complexity of automation with its simple hierarchy (Jamieson & Skraaning, 2018). Furthermore, our analysis reveals inconsistencies in automation research, suggesting potential shortcomings in its effectiveness in communicating automation capabilities.

Drivers’ understanding of automation often diverges from what is offered by SAE; instead of single levels, people have conceptual models containing elements of multiple levels that imply different assumptions about context, vehicle, and driver (Novakazi et al., 2021). Moreover, each level may contain different embodiments of DDT, OEDR, and ODD as offered by manufacturers, offering different features within a defined level. The methods section should have a comprehensive description of the human role in system control. (Kaber, 2018; J. H. Yang et al., 2017). This is particularly relevant for experiments focusing on specific scenarios (e.g., highway driving) with diverse user roles and automation competencies that span multiple SAE levels.

Footnotes

Acknowledgements

Thank you to members of the Cognitive Sciences Lab. This project was supported by the National Science Foundation under grant number 1839484. Any opinions and conclusions expressed within are those of the authors and do not necessarily reflect the views of the National Science Foundation.

ORCID iD

Anthony McDonald

References

Ataya

Kim

Elsharkawy

Kim

(2021). How to Interact with a Fully Autonomous Vehicle: Naturalistic Ways for Drivers to Intervene in the Vehicle System While Performing Non-Driving Related Tasks. Sensors, 21(6), Article 6.

Biondi

F. N.

Lohani

Hopman

Mills

Cooper

J. M.

Strayer

D. L.

(2018). 80 MPH and out-of-the-loop: Effects of real-world semiautomated driving on driver workload and arousal. Proceedings of the HFES Annual Meeting, 62(1), 1878–1882.

Blommer

Curry

Kochhar

Swaminathan

Talamonti

Tijerina

(2015). The Effects of a Scheduled Driver Engagement Strategy in Automated Driving. Proceedings of the HFES Annual Meeting, 59(1), 1681–1685.

Blommer

Curry

Kochhar

Swaminathan

Talamonti

Tijerina

(2017). Off-Road Glance Behavior, Response Time to a Forward Collision Hazard, and Engagement Strategy Effects in Automated Driving. Proceedings of the HFES Annual Meeting, 61(1), 1909–1913.

Choi

J. K.

Y. G.

(2015). Investigating the Importance of Trust on Adopting an Autonomous Vehicle. International Journal of HumanComputer Interaction, 31(10), 692–702.

Clark

Feng

(2016). Age differences in the takeover of vehicle control and engagement in non-driving-related activities in simulated driving with conditional automation. Accident Analysis & Prevention, 106, 468–479.

Clark

McLaughlin

A. C.

Williams

Feng

(2017). Performance in Takeover and Characteristics of Non-driving Related Tasks during Highly Automated Driving in Younger and Older Drivers. Proceedings of the HFES Annual Meeting, 61(1), 37–41.

Cohen-Lazry

Borowsky

Oron-Gilad

(2017). The Effects of Continuous Driving-Related Feedback on Drivers’ Response to Automation Failures. Proceedings of the HFES Annual Meeting, 61(1), 1980–1984.

Coverdale

Roberts

Louie

Beresin

(2006). Writing the Methods. Academic Psychiatry, 30(5), 361–364.

10.

Feldhütter

Gold

Hüger

Bengler

(2016). Trust in Automation as a Matter of Media Influence and Experi-ence of Automated Vehicles. Proceedings of the HFES Annual Meeting, 60(1), 2024–2028.

11.

Funkhouser

Drews

(2016). Putting the Brakes on Autonomous Vehicle Control: Responding to System Breakdowns. Proceedings of the HFES Annual Meeting, 60(1), 1859–1863.

12.

Gasser

T. M.

Westhoff

(2012). BASt-study: Definitions of automation and legal issues in Germany. Proceedings of the 2012 Road Vehicle Automation Workshop.

13.

Glaser

Green

Llaneras

R. E.

Meyer

(2017). Driver Hazard Detection and Avoidance Performance as a Function of EyesOff-Road Interval Under Partially Automated Driving. Proceedings of the HFES Annual Meeting, 61(1), 1914–1918.

14.

Gold

Berisha

Bengler

(2015). Utilization of Drivetime – Performing Non-Driving Related Tasks While Driving Highly Automated. Proceedings of the HFES Annual Meeting, 59(1), 1666–1670.

15.

Gold

Damböck

Lorenz

Bengler

(2013). “Take over!” How long does it take to get the driver back into the loop? Proceedings of the HFES Annual Meeting, 57(1), 1938–1942.

16.

James

(1890). The principles of psychology, Vol I. Henry Holt and Co.

17.

Jamieson

G. A.

Skraaning

(2018). Levels of Automation in Human Factors Models for Automation Design: Why We Might Consider Throwing the Baby Out With the Bathwater. Journal of Cognitive Engineering and Decision Making, 12(1), 42–49.

18.

Kaber

D. B.

(2018). Issues in Human–Automation Interaction Modeling: Presumptive Aspects of Frameworks of Types and Levels of Automation. Journal of Cognitive Engineering and Decision Making, 12(1), 7–24.

19.

Kallet

R. H.

(2004). How to Write the Methods Section of a Research Paper. Respiratory Care, 49(10), 1229–1232.

20.

Kerschbaum

Lorenz

Bengler

(2014). Highly automated driving with a decoupled steering wheel. Proceedings of the HFES Annual Meeting, 58(1), 1686–1690.

21.

Körber

Baseler

Bengler

(2018). Introduction matters: Manipulating trust in automation and reliance in automated driving. Applied Ergonomics, 66, 18–31.

22.

Kotz

Cals

J. W. L.

(2013). Effective writing and publishing scientific papers, part IV: Methods. Journal of Clinical Epidemiology, 66(8), 817.

23.

Lee

See

(2004). Trust in Automation: Designing for Appropriate Reliance. Human Factors, 46(1).

24.

Lee

Green

Seppelt

McGehee

Rau

Johansson

Merat

Balk

Philips

Bengler

Kitazaki

Itoh

Engström

(In Press). Mental Models of Driving Automation: Definition, mechanisms, consequences, and implications. Human Factors

25.

Miller

Johns

Mok

Gowda

Sirkin

Lee

(2016). Behavioral Measurement of Trust in Automation: The Trust Fall. Proceedings of the HFES Annual Meeting, 60(1), 1849–1853.

26.

Miller

Sun

Johns

Ive

Sirkin

Aich

(2015). Distraction Becomes Engagement in Automated Driving. Proceedings of the HFES Annual Meeting, 59(1), 1676–1680.

27.

Miller

E. E.

Boyle

L. N.

(2017). Driver adaptation to lane keeping assistance systems: Do drivers become less vigilant? Proceedings of the HFES Annual Meeting, 61(1), 1934–1938.

28.

Morris

D. M.

Erno

J. M.

Pilcher

J. J.

(2017). Electrodermal Response and Automation Trust during Simulated Self-Driving Car Use. Proceedings of the HFES Annual Meeting, 61(1), 1759–1762.

29.

National Highway Traffic Safety Administration. (2013). Preliminary statement of policy concerning automated vehicles.

30.

Neubauer

Matthews

Saxby

(2014). Fatigue in the Automated Vehicle: Do Games and Conversation Distract or Energize the Driver? Proceedings of the HFES Annual Meeting, 58(1), 2053–2057.

31.

Novakazi

Johansson

Strömberg

Karlsson

(2021). Levels of What?Investigating Drivers’ Understanding of Different Levels of Automation in Vehicles. Journal of Cognitive Engineering and Decision Making, 15(2–3), 116–132.

32.

Radlmayr

Briich

Schmidt

Solbeck

Wehner

(2018). Peripheral monitoring of traffic in conditionally automated driving. Proceedings of the HFES Annual Meeting, 62(1), 1828–1832.

33.

Reagan

I. J.

Kidd

D. G.

Cicchino

J. B.

(2017). Driver Acceptance of Adaptive Cruise Control and Active Lane Keeping in Five Production Vehicles. Proceedings of the HFES Annual Meeting, 61(1), 1949–1953.

34.

SAE International. (2021). SAEJ3016: Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles.

35.

Sanders

Kaplan

Koch

Schwartz

Hancock

P. A.

(2019). The Relationship Between Trust and Use Choice in Human-Robot Interaction. Human Factors, 61(4), 614–626.

36.

Schmidt

Dreißig

Stolzmann

Rötting

(2017). The Influence of Prolonged Conditionally Automated Driving on the Take-Over Ability of the Driver. Proceedings of the HFES Annual Meeting, 61(1), 1974–1978.

37.

Yang

J. H.

Han

Park

J.-M.

(2017). Toward Defining Driving Automation from a Human-Centered Perspective. CHI ‘17 Extended Abstracts, 2248–2254.

38.

Yang

Gerlicher

Bengler

(2018). How does relaxing posture influence take-over performance in an automated vehicle? Proceedings of the HFES Annual Meeting, 62(1), 696–700.