Next Article in Journal
Effect of Digital Transformation in Sports Companies on Green Innovation: Evidence from Listed Companies in China
Previous Article in Journal
Integration of Technological Tools in Teaching Statistics: Innovations in Educational Technology for Sustainable Education
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 16
Issue 19
10.3390/su16198345
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Analysis of Influencing Factors of Level 3 Automated Vehicle Takeover: A Literature Review

by
Hanying Guo
*,
Haoyu Qiu
,
Yongjiang Zhou
and
Yuxin Deng
Department of Automobile and Transportation, Xihua University, Chengdu 610039, China
*
Author to whom correspondence should be addressed.
Sustainability 2024, 16(19), 8345; https://doi.org/10.3390/su16198345
Submission received: 21 August 2024 / Revised: 19 September 2024 / Accepted: 24 September 2024 / Published: 25 September 2024
Browse Figures
Versions Notes

Abstract

:
Level 3 automated vehicles (L3 AVs) enable the driver to perform non-driving tasks, taking over in an emergency. In recent years, studies have extensively discussed the influencing factors of L3 AV takeovers. Extensive literature review shows that L3 AV takeovers are affected by human factors, traffic environment, and automatic driving systems. On this basis, this study proposes a conceptual framework of L3 AV takeovers. The main findings of this study include the following: (1) non-driving tasks, non-driving posture, individual characteristics, and trust have an impact on takeover behavior; (2) high traffic density, poor road geometry, and extreme weather have a negative impact on the takeover; (3) multimodal interaction design can improve collection performance. Although the existing research has made rich achievements, there are still many challenges. The influence of human factors on takeover performance is controversial, the quantification standard of takeover influencing factors is insufficient, and the prediction accuracy needs to be improved. It is suggested to refine the criteria of driver participation in NDRT, formulate an effective measurement standard of driver fatigue, and develop a takeover prediction model combining driver status and traffic environment conditions. It provides a research basis for the formulation of laws, infrastructure construction, and human–computer interaction design.

1. Introduction

In recent years, with the rapid development of autonomous driving technology, the automation level of vehicles has been continuously improved. In 2014, the American Society of Automotive Engineers (SAE) developed a classification standard for the level of vehicle automation, which divided the level of vehicle automation into six levels: No Automation (L0), Driver Assistance (L1), Partial Automation (L2), Conditional Automation (L3), High Automation (L4), and Full Automation (L5) [1]. In April 2021, SAE and the International Organization for Standardization (ISO) updated the class definition for autonomous driving to meet the distinction between “driver assistance systems” and “automated driving system (ADS)” when driver assistance and active safety features are abundant [2]. According to SAE’s definition of an autonomous driving class, the current technology is still in the Level 3 autonomous driving phase.
When the road environment of the L3 AV exceeds the Operational Design Domain (ODD), the vehicle needs to switch from autonomous driving to human driving. However, when the driver participates in NDRT during the driving process of the autonomous vehicle, the driver lacks supervision of the traffic environment and cannot quickly construct the mental representation of the current traffic environment when facing danger, which leads to huge safety risks when the driver takes over the control of the vehicle. In past studies, L3 AV takeover has been widely concerned and discussed by scholars.
The most important process of L3 AV takeover is the driver’s processing and decision-making of external information, including stimulation of sensory organs, cognitive processing of information, decision selection, driving recovery preparation (putting hands and feet back on the steering wheel and pedals), and vehicle manipulation [3]. The driver’s processing of external information is affected by multiple factors, including human factors, traffic environment, human–computer interaction system, etc. ADS plays a decision-making role, dominating vehicle driving through perception and decision-making, but its ODD is limited by the complexity of the traffic environment and the driver’s attention. Traffic environment is an external variable that affects ADS and human factors. The complexity of the traffic environment determines the operating pressure of ADS and drivers and affects the synergistic effect of the two. Among the human factors, the driver is the last line of defense to take over the vehicle. When the ADS cannot cope, the driver needs to take over the vehicle quickly, which depends on whether the ADS provides a timely warning and the driver’s reaction ability. Therefore, in the L3 AV, the three must form a good collaboration to ensure that the vehicle can operate safely and efficiently in different driving environments. These effects can be represented by a conceptual framework, as shown in Figure 1.
In the process of L3 AV, the role of the driver changes from the active operator of the vehicle to the passive monitoring tube, resulting in a significant difference in the status and behavior of the driver compared with the traditional manual driving vehicle [4], which makes the study of the influencing factors of L3 AV takeover crucial. Conducting comprehensive research on the influencing factors of L3 AV is crucial for enhancing their safety, fostering public acceptance, and facilitating market deployment. This study reviews the existing relevant studies and finds that there are inconsistencies in the results of similar studies. The existing quantitative standards on the factors affecting takeover performance are not comprehensive enough, and the prediction accuracy of takeover performance needs to be improved.
The main contributions of this study are as follows: (1) from the aspects of human factors, traffic environment, and ADS, it comprehensively summarizes the relevant studies on the factors affecting L3 AV takeover performance; (2) found that the existing studies on the influencing factors of L3 AV takeover had deficiencies in the experimental design, and prospected the future research directions.

2. Research Methodology

2.1. Research Subjects

Based on the previous research of the team [5], the takeover of L3 automatic driving is divided into three consecutive stages: request and response phase, identification and decision phase, and operation and risk avoidance phase. The schematic diagram of automatic driving takeover is shown in Figure 2 below.
(1) Request and response phase: In the process of automatic driving, when the traffic scene exceeds ODD, ADS sends a takeover request (TOR) to the driver, and the process of the driver receiving TOR and responding is called the request and response phase. The time required to complete this phase is defined as the takeover response time [6]. This phase consists of two sub-stages: (1) receiving takeover prompts from the system and (2) completing the preparations for the takeover [5,6,7].
(2) Identification and decision phase: After receiving the TOR, the driver recognizes the surrounding traffic conditions and makes decisions through situational awareness. This stage can be divided into two sub-stages: road condition recognition and driving decision. Among them, the time of road condition recognition is related to the driver’s trust in the ADS. After recognizing traffic conditions, the driver makes driving decisions through the situational awareness ability [7,8,9].
(3) Operation and risk avoidance phase: After the driver recognizes and makes decisions, the vehicle is controlled to increase time to collision (TTC) and collision distance [10,11] through the braking system to avoid dangerous situations.
According to the above three phases, this study will analyze the influencing factors of L3 AV takeover from three aspects: human factors, traffic environment, and automated driving system.

2.2. Research Methods and Inclusion Criteria

Five databases (Elsevier, Web of Science, Google Scholar China National Knowledge Infrastructure, and Xplore IEEE) were conducted to gather the factors influencing L3 AV takeover. Search keywords for the literature included “automated vehicles”, “takeover”, “takeover request”, and “human-machine interface”. Finally, the literature was filtered using the logical symbols “OR” and “AND” to exclude literature interference not related to the research.
In order to further improve the literature selection, the snowball technique was adopted in this study [12]. Through the forward and reverse snowball method, the literature in the field of L3 AV takeover influence factors is further expanded and comprehensively covered. In addition, specific criteria were included to ensure that the final literature was relevant to the content of the study. The following criteria were established:
  • Articles must involve self-driving takeover;
  • The article should focus on the impact of the driver as the agent, the traffic environment, and the autonomous driving system on the takeover;
  • Articles must be from peer-reviewed journals.
In addition, the following criteria were excluded:
  • Technical reports or official government documents are not included;
  • Papers that focus solely on human-driven vehicles are excluded.
The search policy flowchart, as shown in Figure 3, Outlines the detailed process of searching for a policy.
The number of published papers can not only reflect the scholars’ attention to the discipline but also become an important quantitative index to evaluate the research activities and development trends of a certain field in a specific period [13]. The annual published papers on the influencing factors of L3 AV takeover are shown in Figure 4.
The annual publication volume of literature related to the impact of the L3 AV takeover generally shows an increasing trend, which can be divided into two stages. The first stage is the preliminary development stage (between 2013 and 2018). As a relatively new field, there are few studies on L3 AV and few achievements. The second stage is the rapid development stage (between 2019 and 2023); with the rapid development of autonomous driving technology, the L3 AV takeover has received much attention.

3. Human Factor

When performing driving and non-driving tasks, the driver’s attention and alertness decrease over time [14]. When the driver switches between the driving task and the non-driving task, the takeover performance will be different due to the heterogeneity of the driver. Individual heterogeneity is related to the driver’s physiological factors and the degree of trust in the system [15]. Therefore, the influence of human factors on the takeover process has been widely considered by the academic community and has become one of the main research hotspots in this field.

3.1. Non-Driving-Related Task

L3 AV technology can free up drivers so that they have more free time during driving. This means the driver can engage in some non-driving-related task (NDRT) at the same time.
During autonomous driving, NDRT participation significantly affects the driver’s takeover performance. Zeeb et al.’s [16] study proves this view: when the driver performs NDRT in the active state of automatic driving and the NDRT has visual resource demand (for example, watching video), the driver’s takeover performance is worse than that without a visual driving task. Weaver et al. [17] found that when the NDRT has manual resource demands (e.g., playing a smartphone game), takeover performance is worse than when there are no manual resource demands. The study of Wang et al. [18] also supports this conclusion, when the driver is engaged in NDRT with manual resource demands, after receiving TOR, he needs to first remove his hands from the device for non-driving-related tasks and then put them on the steering wheel, which leads to a longer takeover time.
Interestingly, other scholars have a different view. In the process of automatic driving, NDRT can effectively alleviate the driver’s mental fatigue and improve the takeover performance [19]. Naujoks et al. [20] found that the duration of automatic driving did not affect the driver’s takeover performance. Sleepiness remained relatively low despite considerable periods of autopilot, which may be the result of the driver’s stimulation while participating in NDRT. The study by Miller et al. [21] found that simply monitoring the autopilot system led to more drowsiness than watching videos and reading news. In the event of driver fatigue, appropriate NDRT can maintain some level of situational awareness and increase driver alertness [22,23,24].
When a driver performs an NDRT, it is affected by individual driver differences (age, experience, etc.). Studies have shown that drivers of different ages have the same reaction speed [10], but young drivers are more willing to participate in NDRT than older drivers, and the reaction time of taking over after participating in NDRT is also faster [6,10]. In addition, Liang et al. [25] showed that experienced drivers are better at combining NDRT with driving, which can reduce the negative impact of NDRT on takeover performance.
In addition to individual differences, studies have found that different types of NDRTS have significant differences in takeover performance. Lee et al. [26] compared nine kinds of NDRTs based on the attributes of NDRTs (i.e., the occupation of drivers’ motor, cognitive, and visual resources), and the results showed that drivers’ cognitive load was significantly negatively correlated with both vertical and horizontal control measures. Muller et al. [27] compared NDRTS for reading, listening, and watching movies and found that different NDRTS had significant differences in psychological workload, and NDRTS associated with high workload (such as reading) would lead to longer reaction time.

3.2. Non-Driving Postures

L3 AVs do not require the driver to continuously monitor or control the vehicle, so the driver can conduct NDRT for long periods of time without interference from driving tasks. If most of the driver’s time is no longer spent driving, then the driver may adopt a non-driving position [28] to prevent the onset of motion sickness and improve driving comfort [29,30].
In the process of autonomous driving, the non-driving position has a significant negative impact on takeover performance. Cao et al.’s [31] research showed that when the driver takes over in the non-driving position state, the takeover time is longer, and the takeover performance is worse. However, there are also studies that show that the driver in the non-driving position has no significant impact on the takeover time during automatic driving [32]. At the same time, the non-driving position may cause harm to the performance of current occupant safety systems, such as seat belts and airbags, which were developed for members adopting standard driving positions [33].

3.3. Individual Characteristics

The physiological factors of the driver will have an impact on the identification and decision-making phase of the takeover. Therefore, this section will discuss the effects of the driver’s age, fatigue level, and alcohol on the identification and decision-making stages.

3.3.1. Age

Age is a major factor affecting cognitive function, which declines with age [34]. Driving is a complex task, and a decline in cognitive function can easily lead to poor driving decisions that are detrimental to road safety.
During autonomous driving, the driver’s participation in NDRT can have an impact on takeover safety, resulting in more collisions during takeover and the vehicle deviating from the lane center line [16,35]. Under high-intensity NDRT, older driver takeovers showed lower accuracy and longer takeovers [36]. Research by Gong et al. [37] also showed that older drivers had longer takeover time and lower lateral control stability than younger drivers.
Clark et al. [38] found that older drivers drove slower than younger drivers, deviated less from the center of the lane, and were more inclined to brake when taking over. Korber et al. [10] found that the driving strategies adopted by drivers of different ages were different, but the reaction speed was the same. Older drivers brake more frequently and harder in order to maintain a longer minimum TTC; younger drivers showed stronger braking only in difficult traffic situations.

3.3.2. Fatigue

Fatigue is extreme tiredness caused by mental or physical exertion or illness [39].
Fatigue is another risk factor that affects takeover. During autopilot, the monotonous driving environment can easily induce fatigue, leading to drowsiness or a slow loss of alertness [40,41]. Fatigue will increase over time [42], leaving the driver with no control over the vehicle when he or she takes over.
Goncalves et al.’s [43] study found that most subjects reported high fatigue before the 15 min automatic driving. Such self-perceived fatigue enabled subjects to reduce their lateral control of the vehicle during the takeover period, thus leading to impaired takeover performance. Jarosch et al.’s [44] study also showed that prolonged L3 autopilot while engaged in monitoring tasks (i.e., monitoring the autopilot system) would lead to fatigue and have a negative impact on takeover performance.
However, a small number of studies have found inconsistent results. For example, Vogelpohl et al. [45] found that driver fatigue did not significantly lead to longer takeover times. At the same time, Hirsc et al. [46] confirmed that takeover time and standard difference in self-drive speed were not significantly affected by driver fatigue.
The current research is unable to accurately quantify the driver fatigue state, leading to some controversial conclusions. In order to effectively analyze the effect of fatigue on takeover performance, a more refined method for quantifying fatigue degree needs to be introduced.

3.3.3. Alcohol

Drunk driving has been a serious road safety issue throughout history [47].
In Level 3 autonomous driving tasks, alcohol can blur the driver’s consciousness, which can negatively affect taking over. Wiedemann et al. [48] studied the effect of alcohol on drivers’ takeover behavior and found that alcohol would increase takeover time and reduce takeover performance. Participation in NDRT may lead to more serious accident consequences if alcohol is consumed [49].
In addition, differences in individual physical fitness can lead to differences in takeover performance after drinking. Due to the difference in body metabolism, the concentration of body fluids of different drivers who drink the same amount of alcohol will be different, and this difference will lead to the deviation of drivers’ takeover performance [50].

3.4. Trust

In the field of automation, trust is defined as the attitude that an agent will help achieve personal goals in situations of uncertainty and instability [51]. In the field of autonomous driving, trust is the driver’s subjective evaluation of the trustworthiness of the automated driving system and largely determines whether and how the driver will use the automated driving system. This section will explore the impact of trust on takeovers from both the perspective of trust establishment and maintenance.

3.4.1. Establishment of Trust

The establishment of trust is affected by individual driver differences such as age, culture, gender, and personality [52]. Zhao et al. [53] found that young people (18–35 years old) have lower trust in ADS than middle-aged people (35–60 years old) and older people (over 60 years old) due to their in-depth understanding of automatic driving technology. The research of Du et al. [54] shows that drivers with a higher background culture have higher trust and preference for ADS. In addition, the establishment of trust not only affects the degree of trust but also affects the driver’s willingness to use ADS. Studies have found that women’s trust in ADS is lower than that of men [55,56], resulting in women’s willingness to use ADS is lower than that of men [56].
Personality plays an important intermediary role in the establishment of trust in autonomous driving. Driver trust in ADS is established through a series of complex psychological processes, and personality plays an indispensable role in trust establishment [57]. Some studies have found that when the driver’s personality characteristics tend to be responsible, the more difficult it is to establish trust with ADS. On the contrary, drivers who tend to be extroverted and open tend to accept and build trust more easily [58]. This may be because responsible personalities are associated with qualities such as order and conscientiousness, and they prefer to be in orderly environments [59].

3.4.2. Impact of Trust

The driver’s excessive trust in the automated driving system will lead to the misuse of the automated driving system and the decline of the monitoring ability [51,60], which will lead to the driver’s wrong judgment on key events and increase the takeover reaction time [61,62], thus leading to accidents [63].
The driver’s subjective trust has a significant impact on the takeover time [15]. Excessive trust in ADS causes drivers to pay more attention to the NDRT and less attention to changes in the surrounding environment [15]. As the driver’s trust in the system increases, they develop the illusion that the system will give them enough time to make decisions and maneuver, resulting in longer takeovers [64,65,66].
Excessive trust can affect the driver’s situational awareness to some extent, allowing for errors of judgment and an increase in takeover time during the takeover. Excessive trust can cause the driver to monitor less of the surrounding environment and reduce the speed at which situational awareness recovers [67].
When ADS fails, different error types may have different effects on trust. Azevedo-Sa et al. [68] analyzed the impact of error types (false positives and false negatives) on the development of driver trust under different road shapes and found that false negatives are more harmful to trust than false positives, and this influence will be strengthened under dangerous road conditions.
Trust is evolving in the driver’s interaction with ADS. Manchon et al. [69] showed that the initial trust level of drivers would affect their behavior. The trust group participated in NDRT more frequently and spent more time monitoring the road than the distrust group, but the trust level of both groups increased in the experiment, indicating that trust gradually evolved during the driving process.

4. Traffic Environment

The impact of the traffic environment on driver takeovers falls into three categories: traffic density, road alignment, and weather. In this section, we will focus our discussion on these three aspects.

4.1. Traffic Density

Traffic density affects the driver’s takeover decision and, thus, the minimum collision time when taking over. In the high-density traffic environment, the smaller the spacing between the two workshops, the shorter the minimum collision time, and the minimum collision time increases by about 40% compared with the low-density traffic environment, which makes the probability of collision higher to some extent [70]. In a complex traffic environment, drivers will selectively ignore the surrounding information and adopt conservative driving strategies [10,71,72]. At the same time, with the increase in traffic density or the decrease in available escape paths, the takeover quality of drivers also decreases [73].
The occupation of escape routes by high-density traffic will reduce takeover performance. Studies have shown that excessive traffic volume will reduce driver takeover performance [10,70,74]. Another study by Du et al. [71] found that oncoming traffic flow did not affect takeover time, but when the escape route was in a high-density traffic environment, the driver’s takeover performance would be significantly reduced.
However, the effect of traffic density on takeover performance depends not only on the external environment but also on the state of the driver when performing the NDRT. Some studies have found that when drivers execute NDRT if the cognitive load of drivers is low, they will pay more attention to the driving environment and respond to the takeover request faster under the condition of high-density traffic flow [71]. However, some scholars have found that there is no significant interaction between NDRT and traffic density [70].

4.2. Route Shape and Weather Conditions

One of the main goals of geometric design is to connect the different road elements and meet the driver’s expectations for safe driving. However, unreasonable geometric design may go against the driver’s expectations and easily lead to dangerous accidents.
Geometric design inconsistencies increase the driver’s cognitive load and prolong takeover time. Borojeni et al. [75] found a strong correlation between road geometry design and TOR urgency under near-realistic simulated conditions. When a takeover request is made on a straight road, drivers respond 5.8% faster to high emergency TOR than to low emergency TOR. On winding roads, however, drivers responded 7.4 percent slower to high emergency TOR than low emergency TOR.
When facing a curve of high curvature, the driver will choose a more conservative takeover strategy to ensure the safety of the takeover. In the process of takeover, when the driver enters the curve at a high speed, the visual distance will be reduced if the curvature is too high, and the driver will brake and slow down suddenly to ensure the safe passage of the curve [76]. When the driver receives TOR in a corner and asks to take over, the driver will often apply stronger brakes to ensure safe driving, which will result in greater lateral deflection of the vehicle [77].
Extreme weather will affect the driver’s mental representation of the construction of the traffic environment and have a negative impact on the takeover. In extreme environments such as rain, snow, and fog, the driver’s situational awareness will be affected, making it unable to quickly identify hazards and take over [78]. In foggy weather, when the fog concentration is high, the driver’s sight moves between the road ahead and the dashboard, while when the fog is low, the driver’s gaze is almost centered on the center of the road, possibly to better help the driver regain situational awareness and allow the driver to better take over [79].

5. Automated Driving System

ADS is an advanced electronic safety system that is used to replace the driver to execute dynamic driving tasks (DDTs) during the driving process, But the limitations of the technology make the current ADS imperfect, and the driver needs to take over the vehicle when the ODD is exceeded. This section will discuss two aspects of human–computer interaction and the transfer mode of control.

5.1. Human–Computer Interaction

The interactive system is the hub of information exchange between the driver and the ADS, aiming to complete the defined tasks through certain interactions. When the ADS detects a danger, it alerts the driver and sends out TOR through the interactive system, which could lead to an accident if the driver fails to take over in time. A good interactive system can effectively guide the driver to take over the vehicle and avoid accidents. The existing research generally focuses on vision, hearing, touch, smell, multimodal hybrid interaction, and so on. In this section, these aspects are discussed and analyzed.

5.1.1. Visual Interaction

A visual interaction system can provide intuitive, usable information to the driver when the driver is faced with takeover. The interface of visual interaction is mostly located near the dashboard or middle screen, and icons are generally used to remind the driver to take over the vehicle [80].
A single visual interaction will prolong the takeover time and increase the risk of accidents [81,82,83]. The driver needs to see the TOR emitted by the visual interaction in order to react, but the TOR of a single visual interaction may be subconsciously ignored. Therefore, most studies do not recommend the use of a single visual interaction, and the average response time of a single visual interaction is 2.82 s, which is higher than other single-mode interaction mode times [80].
Wraparound visual interaction can effectively attract the driver’s attention and improve takeover performance. Meschtscherjakov et al. [84] designed a light surround experiment in which the dynamic changes of the surround visual interaction better captured the driver’s attention. Yang et al. [85] designed an LED light located under the windshield and found that changing the color, frequency, and lighting position of the LED could effectively help the driver regain situational awareness and improve takeover performance. Other studies have shown that visual interaction can convey a large amount of information and can be directly perceived by vision without considering the driver’s visual focus [80,86].

5.1.2. Auditory Interaction

Auditory interaction is a way that humans interact with the external environment through the auditory system, and its properties include semantics, speed, pitch, and timbre.
The dangerous atmosphere brought by semantics can shorten the driver’s takeover time. Studies have shown that words that represent a crisis are more likely to indicate an impending danger than those that announce it, allowing the driver to react faster and take over [87,88]. Politis et al. [88] found that for highly urgent speech messages, the response time for a single auditory interaction was 2.24 s, significantly lower than the response time for two modes of interaction, single touch, and single vision. In terms of accuracy, the auditory interaction was 94%, the tactile interaction was 88%, and the visual interaction was only 40%, with the auditory interaction taking over more security.
The information conveyed by an unreasonably high speed of speech may not be accepted by the driver, leading to the failure of the takeover. If the driver is not satisfied with the information received, he is likely to ignore the signal, which may lead to more serious consequences [89].
An emotive tone of voice grabs the driver’s attention better. A sharp tone with strong emotions creates an atmosphere of tension and urgency for the driver, allowing the driver’s attention to catch the information, take over the vehicle more quickly [90], and reduce the probability of an accident.
Drivers were more receptive to female voices, which they thought were softer and more pleasant, but male voices helped take over performance better [89,91]. Research by Richie et al. [92] showed that when drivers received TOR, the response time for male voices was 1.16 s, compared to 1.45 s for female voices, with the male voice 0.29 s faster than the female voice. This may be because people are more submissive to authoritative voices since male voices are generally perceived as more authoritative than female voices.

5.1.3. Tactile Interaction

Tactile interaction uses vibrations to stimulate different parts of the driver’s torso to achieve different types of takeover commands. Tactile interaction is often designed for areas that are in frequent contact with the driver, such as the steering wheel, seats, etc.
Borojeni et al. [93] designed a deformable vibration steering wheel that uses the vibration and deformation of the steering wheel to alert the driver to take over, which can effectively reduce the workload of the driver.
The tactile interaction of the seat can effectively improve the driver’s takeover time. In the process of automatic driving, the driver is often in contact with the seat, so the tactile interaction of the seat can serve as a good reminder [94]. When tactile interaction provides clear directional information through vibration, it can reduce the driver’s takeover time and improve takeover performance [95]. This finding was also confirmed by Cohen-Lazry et al. [96], who set six tactile receptors on the seat cushion and used pulse signals to remind the driver. When the pulse signal opposes the direction of danger, the driver’s reaction time is 0.92 s. Conversely, When the pulse signal matches the direction of danger, the driver’s reaction time increases to 1.12 s, highlighting that when the pulse direction is the same as the direction of danger avoidance, the driver’s reaction speed is faster.

5.1.4. Olfactory Interaction

The human sense of smell can be used not only to enjoy food but also to sense environmental hazards [97]. As a result, some scholars have conducted research on olfactory interactions.
Tang et al. [98] found that the pungent smell of minty can effectively increase the level of driver alertness. However, compared with auditory and tactile interaction, the stimulation of olfactory interaction was not obvious, and the presence or absence of minty odor had no significant effect on takeover time. Meng et al.’s [99] study showed that olfactory stimulation was not suitable for time-tight takeover scenarios because olfactory stimulation could not occur in time compared with auditory and tactile stimulation. Bodnar et al.’s [100] study also showed that without specialized odor recognition training, it is difficult to distinguish similar odors.
Each type of interaction has its own advantages and disadvantages. The advantages and disadvantages of each type of single-modal interaction are shown in Table 1 below.

5.1.5. Multimodal Interaction Modes

In order to overcome the defects of single mode and ensure the effective transmission of takeover request information to the driver, multi-mode takeover requests began to appear, such as visual–auditory dual mode, auditory–tactile dual mode, visual–tactile dual mode, and auditory–visual–tactile three modes.
Naujoks et al. [81] found that the auditory–visual dual mode is more conducive to increasing lateral control stability than the single visual mode. Petermeijer et al.’s [102] study showed that the audiory–tactile dual mode was more conducive to improving the takeover response time than the single tactile and single auditory mode. Politis et al. [103] showed that the takeover response time of the visual–tactile dual mode and the auditory–visual–tactile three modes were shorter, and the lateral control stability was higher than that of the single visual mode, but there was no significant difference in the takeover response time and lateral control stability between the auditory–visual–tactile three modes and the visual–tactile dual mode.

5.2. Control Transfer Mode

In the research on control transfer mode, there are two common modes: one is share control, and the other is trade control.
Share control is when the driver and the automated driving system engage in the driving task at the same time [104]. Li et al. [105] designed a novel haptic takeover controller to achieve smooth control switching by estimating the expected path of the driver and the automated system to obtain optimal steering inputs, generating haptic guidance torque to assist the driver in regaining control. In extreme cases, the driver needs to switch control in a curve, resulting in a sudden interruption of ADS, which will cause the vehicle to steer unsmoothly. In this case, Okada et al. [106] developed a method of sharing control through haptics to ensure the stability of the vehicle and the steering.
Trade control refers to the alternate handling of driving tasks by the driver and the automated driving system [107]. A representative model of trade control is a two-stage early warning. The driver regains some situational awareness in the interval between the two alerts, knowing the current traffic conditions ahead of time and reducing the pressure on the driver to take over [108]. Zhang et al. [109] studied the time interval between the two warnings and found that the driver restored a high level of situational awareness within the 5 s interval.

6. Discussion

This study summarizes the research fields of influencing factors of L3 AV takeover and identifies three main areas that affect L3 AV takeover: human factors, traffic environment, and automated driving system. The three main influencing factors are discussed, and their influence on the takeover is analyzed.

6.1. Human Factors

In the takeover process, the influence of human factors is mainly reflected in the driver’s driving strategy, cognitive load, situational awareness, trust, and other aspects.
There is a certain correlation between the driver’s driving strategy and the driver’s age. When taking over, young drivers will adopt aggressive driving strategies, while older drivers prefer more conservative driving strategies such as braking.
Age and NDRT affect a driver’s cognitive load. When the driver’s cognitive load is overloaded, it can lead to prolonged takeover time and increase the risk of takeover failure. Younger drivers have shorter reaction times and higher accuracy rates at the same cognitive load. Reducing driver participation in high-intensity NDRT helps improve takeover performance.
Fatigue and alcohol can affect a driver’s situational awareness. A single driving environment easily induces driver fatigue, and fatigue causes drivers to lose their perception of the traffic environment. Proper NDRT can help relieve fatigue, restore situational awareness, and improve takeover performance. Alcohol will blur the driver’s perception and judgment of the external environment, increase the takeover time, and reduce the takeover quality.
The driver’s trust and dependence on the system will affect takeover performance. According to research, it is easier to build trust in self-driving systems when drivers are contentious and open. However, when trust is established, the driver may over-rely on the system, resulting in prolonged takeover time. At the same time, driver trust changes dynamically with driving time and ADS performance. Therefore, maintaining a reasonable level of trust helps to improve takeover response speed.
The driver’s non-driving position is unavoidable when an autonomous vehicle is in motion. The non-driving posture not only directly affects the takeover performance but also indirectly improves the takeover performance by improving the driver’s comfort and reducing driving fatigue.
At present, there are some controversies about the influence of human factors on takeover performance. In different studies, the degree of driver participation in NDRT is different, and the quantitative classification of NDRT is lacking. There is a lack of means to quantify the degree of driver fatigue. If the degree of driver fatigue cannot be identified efficiently and accurately, the impact of fatigue on takeover performance cannot be effectively evaluated. The complex relationship among non-driving posture, NDRT, fatigue, and takeover performance is not fully considered, and there is lack of in-depth research on the mechanism of driver personality traits’ influence on trust.

6.2. Traffic Environment

The complexity of the traffic environment increases the difficulty and risk of takeover. Studies have shown that high traffic density will reduce drivers’ ability to obtain external information, increase psychological stress, and make drivers tend to adopt conservative strategies, intentionally ignoring some information to ensure the success of takeover. The effect of traffic density on takeovers is also related to the state of the driver when performing the NDRT, but whether there is a significant effect is still debated, which may be related to differences in NDRT design and driving state.
The main effect of road alignment on takeover is the reduction in visual distance. The curved road alignment reduces the visible distance to the driver, resulting in stronger braking and longer reaction times when the driver faces the curved TOR. In near-real simulated conditions, drivers responded faster to high emergency TOR on straights than to low emergency TOR, but the reverse was true on curved roads. In extreme weather conditions, however, the drivers’ situational awareness was disturbed, and they could not quickly construct a mental representation of the external traffic environment. As a result, drivers need more time to detect and process external hazards.
Most of the existing studies are carried out under the driving simulator, ignoring the influence of the real environment on the driver. In the scenario of experimental design, the driver’s driving operation is pre-set, which may go against the driver’s own will and habits and, thus, have an uncertain impact on the experimental results.

6.3. Automated Driving System

In the takeover process of automatic driving, both single-mode interaction design and multi-mode interaction design have their own advantages and disadvantages.
In the visual interaction, surround visual interaction can effectively attract the driver’s attention, help quickly perceive the situation, and improve the takeover performance. Although visual interaction can convey a large amount of information, a single visual interaction will result in a longer takeover time.
During auditory interaction, hazard semantic cues allow the driver to take control of the vehicle more quickly. Drivers responded faster to male voices than female ones when responding to TOR, possibly because male voices are more authoritative.
In tactile interaction, information is delivered quickly but in a single form, unsuitable as the primary means of taking over the rest.
Olfactory interaction may increase the driver’s level of arousal, but it is only suitable as an auxiliary stimulus.
Multimodal interaction design can make up for the shortcomings of single-modal interaction design and ensure the accuracy of transmitted information. Most studies show that multimodal interaction design can achieve faster and higher quality driver takeovers than single-modal interaction design. However, there is no significant difference in takeover response time between the three-modal interaction model and the two-modal interaction model. Multimodal interaction mode is more helpful in improving takeover performance than single-modal interaction mode, but the increase in the number of modes will lead to cost, interference with the driver’s takeover decision, and increase the workload. Therefore, a reasonable takeover request interaction mode is needed, rather than the pursuit of a multimodal interaction mode.
The mode of control transfer is crucial to the security of takeover. Two common modes of control transfer are trade control and share control. The phased early warning mode uses the time gap between stages to restore the driver’s situational awareness, thereby improving takeover performance and ensuring a safe takeover. Transaction control reduces the complexity of the simultaneous operation of the driver and the autopilot system, but in emergency situations, the driver can take over the situation where there may be a delay in response, thus reducing the takeover performance. In shared control, the driver and the autopilot system jointly handle driving tasks, improving safety and driving experience, while the transfer of control is more natural. Shared control modes are typically implemented through tactile vibrations that gradually give control to the vehicle and guide the driver to regain control, enabling a smooth transfer of control and ensuring a safe takeover. However, under tactile shared control conditions, the autopilot system does not turn off, causing the driver to experience torque on the steering wheel, which reduces takeover performance. Therefore, the choice of control transfer mode depends on the specific application scenario and driving system design objectives. Using a wide range of measurement methods, it is crucial to empirically evaluate shared control and transaction control from an objective point of view.
The above studies show that L3 AV takeover has numerous influencing factors and complex processes. Future studies should consider human factors such as the driver’s driving strategy, cognitive load, and trust, the complexity of the traffic environment, as well as the multimodal interaction of the automated driving system and the transfer mode of control so as to further improve the takeover performance.

7. Conclusions

Research on autonomous vehicle takeovers reveals different factors that influence takeovers, namely human factors, traffic environment, and autonomous driving systems. Judging from the literature review, these factors affect the performance of driver takeovers in different ways. This review emphasizes the interaction between the driver and the factors. Through the review of the relevant literature, the results are analyzed as follows:
  • Non-driving posture not only directly affects takeover performance but also indirectly improves takeover performance by improving comfort and reducing fatigue;
  • Drivers’ physiological factors affect takeover behavior, which can be reduced by participating in NDRT;
  • Unreasonable trust in ADS can increase takeover time and increase accident risk;
  • In the road environment, factors that affect the safety of an automated takeover include traffic density, geometric alignment of the road, and extreme weather;
  • In ADS, multi-mode interaction design can effectively improve takeover performance, but unreasonable interaction design will hinder drivers from taking over.
Although a large amount of the literature has studied the influencing factors of L3 AV takeover from different aspects, there are still some controversial conclusions. The existing quantitative standards for the influencing factors of L3 AV performance are not comprehensive enough, the prediction accuracy of takeover performance needs to be improved, and the influencing factors of Level 3 automated driving need to be further improved. Suggestions for future research directions are as follows:
  • As the influence of human factors on takeover performance is controversial, the criteria for driver participation in NDRT can be refined in the future, effective driver fatigue measurement standards can be developed, and the complex relationship between NDRT, non-driving postures, and fatigue on takeover performance can be further discussed, which will help to fully understand the influence of human factors on takeover;
  • As driver trust changes dynamically, continuous research can be carried out to analyze the dynamic process of trust level change over time through long-term tracking of the interaction between the driver and automatic driving system so as to reveal the long-term impact of the driver’s dependence on system and trust under different driving environments, and provide a reliable basis for ADS design;
  • Due to current regulations and driver safety issues, most of the studies discussed in this study were conducted on driving simulators with different real feelings, and it is suggested that future studies collect experimental data on real platforms;
  • Based on the limitations of the present study, takeover prediction models combining driver status and traffic environment conditions can be developed in the future. Such models can detect and analyze the driver’s physiological state (such as heart rate, eye movement, fatigue, etc.) as well as traffic and weather conditions in real time to more accurately predict takeover behavior, thereby improving the takeover performance and safety of ADS in different driving scenarios.

Author Contributions

H.G.: guide, writing—review and editing, methodology, funding acquisition. H.Q.: writing—original draft preparation, methodology, software. Y.Z.: writing—original draft preparation, methodology, software. Y.D.: writing—original draft preparation, methodology. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the project of the Sichuan Natural Science Foundation, and the project number is 2023NSFSC0386.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

No data were used for the research described in the article.

Acknowledgments

We gratefully acknowledge the support of the Department of Automobile and Transportation, Xihua University, for this research.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. SAE J 3016; Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems. SAE Internatinoal: Warrendale, PA, USA, 2014.
  2. SAE J 3016; Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles. SAE Internatinoal: Warrendale, PA, USA, 2021.
  3. Zhang, B.; de Winter, J.; Varotto, S.; Happee, R.; Martens, M. Determinants of take-over time from automated driving: A meta-analysis of 129 studies. Transp. Res. Part F Traffic Psychol. Behav. 2019, 64, 285–307. [Google Scholar] [CrossRef]
  4. Merat, N.; Jamson, A.H.; Lai, F.C.H.; Carsten, O. Highly Automated Driving, Secondary Task Performance, and Driver State. Hum. Factors 2012, 54, 762–771. [Google Scholar] [CrossRef] [PubMed]
  5. Guo, H.; Zhang, Y.; Cai, S.; Chen, X.; Galante, F. Effects of Level 3 Automated Vehicle Drivers’ Fatigue on Their Take-Over Behaviour: A Literature Review. J. Adv. Transp. 2021, 2021, 8632685. [Google Scholar] [CrossRef]
  6. Wu, Y.; Kihara, K.; Hasegawa, K.; Takeda, Y.; Sato, T.; Akamatsu, M.; Kitazaki, S. Age-related differences in effects of non-driving related tasks on takeover performance in automated driving. J. Saf. Res. 2020, 72, 231–238. [Google Scholar] [CrossRef] [PubMed]
  7. Depner, N.; Bavendiek, J.; Kn, S.; Uhlig, M. HMI Concept for Safe Driving Mode Transitions. ATZ Worldw. 2020, 122, 26–31. [Google Scholar] [CrossRef]
  8. Du, N.; Yang, X.J.; Zhou, F. Psychophysiological responses to takeover requests in conditionally automated driving. Accid Anal. Prev. 2020, 148, 105804. [Google Scholar] [CrossRef]
  9. Lu, Z.; Coster, X.; de Winter, J. How much time do drivers need to obtain situation awareness? A laboratory-based study of automated driving. Appl. Erg. 2017, 60, 293–304. [Google Scholar] [CrossRef]
  10. Körber, M.; Gold, C.; Lechner, D.; Bengler, K. The influence of age on the take-over of vehicle control in highly automated driving. Transp. Res. Part F Traffic Psychol. Behav. 2016, 39, 19–32. [Google Scholar] [CrossRef]
  11. Yoon, S.H.; Ji, Y.G. Non-driving-related tasks, workload, and takeover performance in highly automated driving contexts. Transp. Res. Part F Traffic Psychol. Behav. 2019, 60, 620–631. [Google Scholar] [CrossRef]
  12. Wohlin, C. Guidelines for snowballing in systematic literature studies and a replication in software engineering. In Proceedings of the 18th International Conference on Evaluation and Assessment in Software Engineering, London, UK, 13–14 May 2014. [Google Scholar] [CrossRef]
  13. Zeng, L.; Li, H.; Lin, L.; Hu, D.J.J.; Liu, H. ESG Standards in China: Bibliometric Analysis, Development Status Research, and Future Research Directions. Sustainability 2024, 16, 7134. [Google Scholar] [CrossRef]
  14. Warm, J.S.; Parasuraman, R.; Matthews, G. Vigilance requires hard mental work and is stressful. Hum. Factors 2008, 50, 433–441. [Google Scholar] [CrossRef] [PubMed]
  15. Jin, M.; Lu, G.; Chen, F.; Shi, X.; Tan, H.; Zhai, J. Modeling takeover behavior in level 3 automated driving via a structural equation model: Considering the mediating role of trust. Accid. Anal. Prev. 2021, 157, 106156. [Google Scholar] [CrossRef] [PubMed]
  16. Zeeb, K.; Buchner, A.; Schrauf, M. Is take-over time all that matters? The impact of visual-cognitive load on driver take-over quality after conditionally automated driving. Accid. Anal. Prev. 2016, 92, 230–239. [Google Scholar] [CrossRef] [PubMed]
  17. Weaver, B.W.; DeLucia, P.R. A Systematic Review and Meta-Analysis of Takeover Performance During Conditionally Automated Driving. Hum. Factors 2022, 64, 1227–1260. [Google Scholar] [CrossRef]
  18. Wang, W.; Li, Q.; Zeng, C.; Li, G.; Zhang, J.; Li, S.; Chen, B. Review of Take-over Performance of Automated Driving: Influencing Factors, Models, and Evaluation Methods. China J. Highw. Transp. 2023, 36, 202–224. [Google Scholar] [CrossRef]
  19. Juan, Y. Effects of Non-Driving Related Task Participation Mode on Takeover Performance of Young Drivers in Prolonged Automated Driving. Master’s Thesis, Chang’an University, Xi’an, China, 2021. Available online: https://link.cnki.net/doi/10.26976/d.cnki.gchau.2021.000487 (accessed on 3 January 2024).
  20. Naujoks, F.; Höfling, S.; Purucker, C.; Zeeb, K. From partial and high automation to manual driving: Relationship between non-driving related tasks, drowsiness and take-over performance. Accid. Anal. Prev. 2018, 121, 28–42. [Google Scholar] [CrossRef]
  21. Miller, D.; Sun, A.; Johns, M.; Ive, H.; Sirkin, D.; Aich, S.; Ju, W. Distraction Becomes Engagement in Automated Driving. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 2015, 59, 1676–1680. [Google Scholar] [CrossRef]
  22. Schömig, N.; Hargutt, V.; Neukum, A.; Petermann-Stock, I.; Othersen, I. The Interaction Between Highly Automated Driving and the Development of Drowsiness. Procedia Manuf. 2015, 3, 6652–6659. [Google Scholar] [CrossRef]
  23. Biondi, F.N.; Lohani, M.; Hopman, R.; Mills, S.; Cooper, J.M.; Strayer, D.L. 80 MPH and out-of-the-loop: Effects of real-world semi-automated driving on driver workload and arousal. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 2018, 62, 1878–1882. [Google Scholar] [CrossRef]
  24. Lie, G.; Linlin, X.; Zengke, Q.; Xu, W. Analysis and Overview of Influencing Factors on Autonomous Driving Takeover. Transp. Syst. Eng. Inf. 2022, 22, 72–90. [Google Scholar] [CrossRef]
  25. Liang, N.; Lim, C.; Yu, D.; Prakah-Asante, K.O.; Pitts, B.J. Predicting Automated Vehicle Takeover Decision During the Nighttime. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 2023, 67, 914–919. [Google Scholar] [CrossRef]
  26. Lee, S.C.; Yoon, S.H.; Ji, Y.G. Effects of Non-Driving-Related Task Attributes on Takeover Quality in Automated Vehicles. Int. J. Hum.–Comput. Interact. 2021, 37, 211–219. [Google Scholar] [CrossRef]
  27. Müller, A.L.; Fernandes-Estrela, N.; Hetfleisch, R.; Zecha, L.; Abendroth, B. Effects of non-driving related tasks on mental workload and take-over times during conditional automated driving. Eur. Transp. Res. Rev. 2021, 13, 15. [Google Scholar] [CrossRef]
  28. Yang, Y.; Klinkner, J.N.; Bengler, K. How Will the Driver Sit in an Automated Vehicle?—The Qualitative and Quantitative Descriptions of Non-Driving Postures (NDPs) When Non-Driving-Related-Tasks (NDRTs) Are Conducted. In Proceedings of the 20th Congress of the International Ergonomics Association (IEA 2018); Bagnara, S., Tartaglia, R., Albolino, S., Alexander, T., Fujita, Y., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 409–420. [Google Scholar]
  29. Bohrmann, D.; Bengler, K. Reclined Posture for Enabling Autonomous Driving. In Human Systems Engineering and Design II; Ahram, T., Karwowski, W., Pickl, S., Taiar, R., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 169–175. [Google Scholar]
  30. Qingyang, H.; Xiaoping, J.; Yikang, Z. Analysis of Drivers’ Driving Posture Change Rule Under the Condition of Automatic Driving Level Improvement. Automot. Eng. 2023, 45, 382–392. [Google Scholar] [CrossRef]
  31. Cao, S.; Tang, P.Y.; Sun, X. Driver Take-Over Reaction in Autonomous Vehicles with Rotatable Seats. Safety 2020, 6, 12. [Google Scholar] [CrossRef]
  32. Yang, Y.; Gerlicher, M.; Bengler, K. How does relaxing posture influence take-over performance in an automated vehicle? Proc. Hum. Factors Ergon. Soc. Annu. Meet. 2018, 62, 696–700. [Google Scholar] [CrossRef]
  33. Dissanaike, S.; Kaufman, R.; Mack, C.D.; Mock, C.; Bulger, E. The effect of reclined seats on mortality in motor vehicle collisions. J. Trauma-Inj. Infect. Crit. Care 2008, 64, 614–619. [Google Scholar] [CrossRef]
  34. Salthouse, T.A. When does age-related cognitive decline begin? Neurobiol. Aging 2009, 30, 507–514. [Google Scholar] [CrossRef] [PubMed]
  35. Radlmayr, J.; Gold, C.; Lorenz, L.; Farid, M.; Bengler, K. How Traffic Situations and Non-Driving Related Tasks Affect the Take-Over Quality in Highly Automated Driving. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 2014, 58, 2063–2067. [Google Scholar] [CrossRef]
  36. Peng, Q.; Wu, Y.; Qie, N.; Iwaki, S. Age-related effects of executive function on takeover performance in automated driving. Sci. Rep. 2022, 12, 5410. [Google Scholar] [CrossRef]
  37. Gong, J.; Guo, X.; Pan, L.; Qi, C.; Wang, Y. Impact of Age on Takeover Behavior in Automated Driving in Complex Traffic Situations: A Case Study of Beijing, China. Sustainability 2022, 14, 483. [Google Scholar] [CrossRef]
  38. Clark, H.; Feng, J. Age differences in the takeover of vehicle control and engagement in non-driving-related activities in simulated driving with conditional automation. Accid Anal. Prev. 2017, 106, 468–479. [Google Scholar] [CrossRef]
  39. Hui, Z.; Ding, N.; Ke, Z.; Naikan, D.; Chaozhong, W. A review of driving fatigue research in human-machine co driving environments. Chin. J. Saf. Sci. 2023, 33, 204–211. [Google Scholar] [CrossRef]
  40. Lal, S.; Craig, A. A critical review of the psychophysiology of driver fatigue—ScienceDirect. Biol. Psychol. 2001, 55, 173–194. [Google Scholar] [CrossRef] [PubMed]
  41. Useche, S.A.; Ortiz, V.G.; Cendales, B.E. Stress-related psychosocial factors at work, fatigue, and risky driving behavior in bus rapid transport (BRT) drivers. Accid Anal. Prev. 2017, 104, 106–114. [Google Scholar] [CrossRef] [PubMed]
  42. Neubauer, C.; Matthews, G.; Langheim, L.; Saxby, D. Fatigue and voluntary utilization of automation in simulated driving. Hum. Factors 2012, 54, 734–746. [Google Scholar] [CrossRef] [PubMed]
  43. Gonçalves, J.; Happee, R.; Bengler, K. Drowsiness in Conditional Automation: Proneness, diagnosis and driving performance effects. In Proceedings of the 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC), Rio de Janeiro, Brazil, 1–4 November 2016; pp. 873–878. [Google Scholar] [CrossRef]
  44. Jarosch, O.; Bellem, H.; Bengler, K. Effects of Task-Induced Fatigue in Prolonged Conditional Automated Driving. Hum. Factors 2019, 61, 1186–1199. [Google Scholar] [CrossRef]
  45. Vogelpohl, T.; Kühn, M.; Hummel, T.; Vollrath, M. Asleep at the automated wheel—Sleepiness and fatigue during highly automated driving. Accid. Anal. Prev. 2019, 126, 70–84. [Google Scholar] [CrossRef]
  46. Hirsch, M.; Diederichs, F.; Widlroither, H.; Graf, R.; Bischoff, S. Sleep and Take-Over in Automated Driving. Int. J. Transp. Sci. Technol. 2019, 9, 42–51. [Google Scholar] [CrossRef]
  47. Christophersen, A.S.; Mørland, J. Drugged driving, a review based on the experience in Norway. Drug Alcohol Depend. 1997, 47, 125–135. [Google Scholar] [CrossRef]
  48. Wiedemann, K.; Naujoks, F.; Worle, J.; Kenntner-Mabiala, R.; Kaussner, Y.; Neukum, A. Effect of different alcohol levels on take-over performance in conditionally automated driving. Accid. Anal. Prev. 2018, 115, 89–97. [Google Scholar] [CrossRef] [PubMed]
  49. Irwin, C.; Iudakhina, E.; Desbrow, B.; McCartney, D. Effects of acute alcohol consumption on measures of simulated driving: A systematic review and meta-analysis. Accid. Anal. Prev. 2017, 102, 248–266. [Google Scholar] [CrossRef]
  50. Veldstra, J.L.; Brookhuis, K.A.; de Waard, D.; Molmans, B.H.W.; Verstraete, A.G.; Skopp, G.; Jantos, R. Effects of alcohol (BAC 0.5 per thousand) and ecstasy (MDMA 100 mg) on simulated driving performance and traffic safety. Psychopharmacology 2012, 222, 377–390. [Google Scholar] [CrossRef]
  51. Lee, J.D.; See, K.A. Trust in automation: Designing for appropriate reliance. Hum. Factors 2004, 46, 50. [Google Scholar] [CrossRef]
  52. Kraus, J.; Scholz, D.; Stiegemeier, D.; Baumann, M. The More You Know: Trust Dynamics and Calibration in Highly Automated Driving and the Effects of Take-Overs, System Malfunction, and System Transparency. Hum. Factors 2020, 62, 718–736. [Google Scholar] [CrossRef] [PubMed]
  53. Zhao, X.; Chen, H.; Li, Z.; Li, H.; Gong, J.; Fu, Q. Influence Characteristics of Automated Driving Takeover Behavior in Different Scenarios. Chin. J. Highw. 2022, 35, 195–214. [Google Scholar] [CrossRef]
  54. Du, N.; Robert, L.; Yang, X.J. Cross-Cultural Investigation of the Effects of Explanations on Drivers’ Trust, Preference, and Anxiety in Highly Automated Vehicles. Transp. Res. Rec. 2023, 2677, 554–561. [Google Scholar] [CrossRef]
  55. Pflugfelder, E.H. Autonomous Vehicles and Gender: A Commentary. Transfers 2018, 8, 104–111. [Google Scholar] [CrossRef]
  56. Liu, P.; Yang, R.; Xu, Z.G. Public Acceptance of Fully Automated Driving: Effects of Social Trust and Risk/Benefit Perceptions. Risk Anal. 2019, 39, 326–341. [Google Scholar] [CrossRef]
  57. Kraus, J.; Scholz, D.; Baumann, M. What’s Driving Me? Exploration and Validation of a Hierarchical Personality Model for Trust in Automated Driving. Hum. Factors 2021, 63, 1076–1105. [Google Scholar] [CrossRef]
  58. Charness, N.; Yoon, J.S.; Souders, D.; Stothart, C.; Yehnert, C. Predictors of Attitudes Toward Autonomous Vehicles: The Roles of Age, Gender, Prior Knowledge, and Personality. Front Psychol. 2018, 9, 2589. [Google Scholar] [CrossRef] [PubMed]
  59. Costa, P.T.; Mccrae, R.R.; Dye, D.A. Facet Scales for Agreeableness and Conscientiousness: A Revision of the NEO Personality Inventory. Personal. Individ. Differ. 1991, 12, 887–898. [Google Scholar] [CrossRef]
  60. Bailey, N.R.; Scerbo, M.W. Automation-induced complacency for monitoring highly reliable systems: The role of task complexity, system experience, and operator trust. Theor. Issues Ergon. Sci. 2007, 8, 321–348. [Google Scholar] [CrossRef]
  61. Mcguirl, J.M.; Sarter, N.B. Supporting Trust Calibration and the Effective Use of Decision Aids by Presenting Dynamic System Confidence Information. Hum. Factors 2006, 48, 656. [Google Scholar] [CrossRef]
  62. Beller, J.; Heesen, M.; Vollrath, M. Improving the driver-automation interaction: An approach using automation uncertainty. Hum Factors 2013, 55, 1130–1141. [Google Scholar] [CrossRef]
  63. Yijing, Z.; Ruosong, C.; Jinfei, M.; Guangbo, D. Construction of Driver Takeover Process and Psychological Model under L3 Level Autonomous Driving Conditions. Psychol. Sci. 2019, 42, 415–421. [Google Scholar] [CrossRef]
  64. Payre, W.; Cestac, J.; Delhomme, P. Fully Automated Driving: Impact of Trust and Practice on Manual Control Recovery. Hum. Factors 2016, 58, 229–241. [Google Scholar] [CrossRef] [PubMed]
  65. Korber, M.; Baseler, E.; Bengler, K. Introduction matters: Manipulating trust in automation and reliance in automated driving. Appl. Erg. 2018, 66, 18–31. [Google Scholar] [CrossRef]
  66. Yingfan, C. Research on the Effect of Driver Trust Matching on Driver Control Tak-Eover. Master’s Thesis, Beijing University of Civil Engineering and Architecture, Beijing, China, 2023. Available online: https://link.cnki.net/doi/10.26943/d.cnki.gbjzc.2023.000469 (accessed on 3 January 2024).
  67. Gawron, V.J. Human Performance, Workload, and Situational Awareness Measures Handbook, 2nd ed.; CRC Press: Boca Raton, FL, USA, 2008. [Google Scholar]
  68. Azevedo-Sa, H.; Jayaraman, S.K.; Esterwood, C.T.; Yang, X.J.; Robert, L.P.; Tilbury, D.M. Comparing the Effects of False Alarms and Misses on Humans’ Trust in (Semi)Autonomous Vehicles. In Proceedings of the Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, Cambridge, UK, 23–26 March 2020. [Google Scholar] [CrossRef]
  69. Manchon, J.B.; Bueno, M.; Navarro, J. How the initial level of trust in automated driving impacts drivers’ behaviour and early trust construction. Transp. Res. Part F Traffic Psychol. Behav. 2022, 86, 281–295. [Google Scholar] [CrossRef]
  70. Gold, C.; Korber, M.; Lechner, D.; Bengler, K. Taking Over Control From Highly Automated Vehicles in Complex Traffic Situations: The Role of Traffic Density. Hum. Factors 2016, 58, 642–652. [Google Scholar] [CrossRef]
  71. Du, N.; Kim, J.; Zhou, F.; Pulver, E.; Tilbury, D.M.; Robert, L.P.; Pradhan, A.K.; Yang, X.J. Evaluating Effects of Cognitive Load, Takeover Request Lead Time, and Traffic Density on Drivers’ Takeover Performance in Conditionally Automated Driving. In Proceedings of the 12th International Conference on Automotive User Interfaces and Interactive Vehicular Applications, Virtual Event, 21–22 September 2020. [Google Scholar]
  72. Doubek, F.; Loosveld, E.; Happee, R.; de Winter, J. Takeover Quality: Assessing the Effects of Time Budget and Traffic Density with the Help of a Trajectory-Planning Method. J. Adv. Transp. 2020, 2020, 6173150. [Google Scholar] [CrossRef]
  73. Chen, F.; Lu, G.; Lin, Q.; Zhang, H.; Ma, S.; Liu, D.; Song, H. Review of drivers’ takeover behavior in conditional automated driving. J. Jilin Univ. (Eng. Ed.) 2024, 1–16. [Google Scholar] [CrossRef]
  74. Gold, C.; Happee, R.; Bengler, K. Modeling take-over performance in level 3 conditionally automated vehicles. Accid. Anal. Prev. 2018, 116, 3–13. [Google Scholar] [CrossRef] [PubMed]
  75. Borojeni, S.S.; Boll, S.C.J.; Heuten, W.; Bülthoff, H.H.; Chuang, L. Feel the Movement: Real Motion Influences Responses to Take-over Requests in Highly Automated Vehicles. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, Montreal, QC, Canada, 21–26 April 2018. [Google Scholar]
  76. Martens, M.; Comte, S.; Kaptein, N. The Effects of Road Design on Speed Behaviour: A Literature Review. Inst. Road Saf. Res. 1997, 36, 1–27. [Google Scholar]
  77. Brandenburg, S.; Chuang, L. Take-over requests during highly automated driving: How should they be presented and under what conditions? Transp. Res. Part F Traffic Psychol. Behav. 2019, 66, 214–225. [Google Scholar] [CrossRef]
  78. Vlakveld, W.; van Nes, N.; de Bruin, J.; Vissers, L.; van der Kroft, M. Situation awareness increases when drivers have more time to take over the wheel in a Level 3 automated car: A simulator study. Transp. Res. Part F Traffic Psychol. Behav. 2018, 58, 917–929. [Google Scholar] [CrossRef]
  79. Louw, T.; Merat, N. Are you in the loop? Using gaze dispersion to understand driver visual attention during vehicle automation. Transp. Res. Part C Emerg. Technol. 2017, 76, 35–50. [Google Scholar] [CrossRef]
  80. Yoon, S.H.; Kim, Y.W.; Ji, Y.G. The effects of takeover request modalities on highly automated car control transitions. Accid. Anal. Prev. 2019, 123, 150–158. [Google Scholar] [CrossRef]
  81. Naujoks, F.; Mai, C.; Neukum, A. The Effect of Urgency of Take-Over Requests During Highly Automated Driving Under Distraction Conditions. Int. Conf. Appl. Hum. Factors Ergon. Ahfe 2014, 7, 431. [Google Scholar]
  82. Baek, S.J.; Yun, H.; Yang, J.H. How do humans respond when automated vehicles request an immediate vehicle control take-over? In Proceedings of the 11th International Conference on Automotive User Interfaces and Interactive Vehicular Applications: Adjunct Proceedings, Utrecht, The Netherlands, 21–25 September 2019. [Google Scholar]
  83. Lee, J.; Yang, J.H. Analysis of Driver’s EEG Given Take-Over Alarm in SAE Level 3 Automated Driving in a Simulated Environment. Int. J. Automot. Technol. 2020, 21, 719–728. [Google Scholar] [CrossRef]
  84. Meschtscherjakov, A.; Döttlinger, C.; Rödel, C.; Tscheligi, M. ChaseLight: Ambient LED Stripes to Control Driving Speed. In Proceedings of the 7th International Conference on Automotive User Interfaces and Interactive Vehicular Applications, Nottingham, UK, 1–3 September 2015. [Google Scholar]
  85. Yang, Y.; Karakaya, B.; Dominioni, G.C.; Kawabe, K.; Bengler, K. An HMI Concept to Improve Driver’s Visual Behavior and Situation Awareness in Automated Vehicle. In Proceedings of the 2018 IEEE International Conference on Intelligent Transportation Systems (ITSC), Maui, HI, USA, 4–7 November 2018. [Google Scholar]
  86. Roche, F.; Brandenburg, S. Should the Urgency of Visual-Tactile Takeover Requests Match the Criticality of Takeover Situations? IEEE Trans. Intell. Veh. 2020, 5, 306–313. [Google Scholar] [CrossRef]
  87. Baldwin, C.L. Verbal collision avoidance messages during simulated driving: Perceived urgency, alerting effectiveness and annoyance. Ergonomics 2011, 54, 328–337. [Google Scholar] [CrossRef] [PubMed]
  88. Politis, I.; Brewster, S.; Pollick, F. Language-based multimodal displays for the handover of control in autonomous cars. In Proceedings of the 7th International Conference on Automotive User Interfaces and Interactive Vehicular Applications, Nottingham, UK, 1–3 September 2015. [Google Scholar]
  89. Bazilinskyy, P.; de Winter, J.C.F. Analyzing crowdsourced ratings of speech-based take-over requests for automated driving. Appl. Erg. 2017, 64, 56–64. [Google Scholar] [CrossRef]
  90. Ljungberg, J.K.; Parmentier, F.B.R.; Hughes, R.W.; Macken, W.J.; Jones, D.M. Listen Out! Behavioural and Subjective Responses to Verbal Warnings. Appl. Cogn. Psychol. 2012, 26, 451–461. [Google Scholar] [CrossRef]
  91. Bazilinskyy, P.; Petermeijer, S.M.; Petrovych, V.; Dodou, D.; de Winter, J.C.F. Take-over requests in highly automated driving: A crowdsourcing survey on auditory, vibrotactile, and visual displays. Transp. Res. Part F Traffic Psychol. Behav. 2018, 56, 82–98. [Google Scholar] [CrossRef]
  92. Richie, E.; Offer-Westort, T.; Shankar, R.; Jeon, M. Auditory Displays for Take-Over in Semi-automated Vehicles. In Digital Human Modeling. Applications in Health, Safety, Ergonomics, and Risk Management; Lecture Notes in Computer Science; Springer: Cham, Switzerland, 2018; Chapter 51; pp. 623–634. [Google Scholar]
  93. Borojeni, S.S.; Wallbaum, T.; Heuten, W.; Boll, S. Comparing Shape-Changing and Vibro-Tactile Steering Wheels for Take-Over Requests in Highly Automated Driving. In Proceedings of the 9th International Conference on Automotive User Interfaces and Interactive Vehicular Applications, Oldenburg, Germany, 24–27 September 2017. [Google Scholar]
  94. Petermeijer, S.M.; Winter, J.C.F.D.; Bengler, K.J. Vibrotactile Displays: A Survey with a View on Highly Automated Driving. IEEE Trans. Intell. Transp. Syst. 2016, 17, 897–907. [Google Scholar] [CrossRef]
  95. Petermeijer, S.M.; Cieler, S.; de Winter, J.C.F. Comparing spatially static and dynamic vibrotactile take-over requests in the driver seat. Accid. Anal. Prev. 2017, 99 Pt A, 218–227. [Google Scholar] [CrossRef]
  96. Cohen-Lazry, G.; Katzman, N.; Borowsky, A.; Oron-Gilad, T. Directional tactile alerts for take-over requests in highly-automated driving. Transp. Res. Part F Traffic Psychol. Behav. 2019, 65, 217–226. [Google Scholar] [CrossRef]
  97. Stevenson, R.J. Olfactory perception, cognition, and dysfunction in humans. Wiley Interdiscip. Rev. Cogn. Sci. 2013, 4, 273–284. [Google Scholar] [CrossRef]
  98. Tang, Q.; Guo, G.; Zhang, Z.; Zhang, B.; Wu, Y. Olfactory Facilitation of Takeover Performance in Highly Automated Driving. Hum. Factors 2021, 63, 553–564. [Google Scholar] [CrossRef]
  99. Meng, F.; Spence, C. Tactile warning signals for in-vehicle systems. Accid. Anal. Prev. 2015, 75, 333–346. [Google Scholar] [CrossRef] [PubMed]
  100. Bodnar, A.; Corbett, R.; Nekrasovski, D. AROMA: Ambient awareness through olfaction in a messaging application. In Proceedings of the 6th International Conference on Multimodal Interfaces, ICMI 2004, State College, PA, USA, 13–15 October 2004. [Google Scholar]
  101. Wintersberger, P.; Dmitrenko, D.; Schartmüller, C.; Frison, A.K.; Maggioni, E.; Obrist, M.; Riener, A. S(C)Entinel-Montoring Automated Vehicles with Olfactory Reliability Displays. In Proceedings of the 24th International Conference on Intelligent User Interfaces, Marina del Ray, CA, USA, 17–20 March 2019. [Google Scholar]
  102. Petermeijer, S.; Bazilinskyy, P.; Bengler, K.; de Winter, J. Take-over again: Investigating multimodal and directional TORs to get the driver back into the loop. Appl. Ergon. 2017, 62, 204–215. [Google Scholar] [CrossRef] [PubMed]
  103. Politis, I.; Brewster, S.; Pollick, F. Using Multimodal Displays to Signify Critical Handovers of Control to Distracted Autonomous Car Drivers. Int. J. Mob. Hum. Comput. Interact. 2017, 9, 1–16. [Google Scholar] [CrossRef]
  104. Marcano, M.; Diaz, S.; Perez, J.; Irigoyen, E. A Review of Shared Control for Automated Vehicles: Theory and Applications. IEEE Trans. Hum. Mach. Syst. 2020, 50, 475–491. [Google Scholar] [CrossRef]
  105. Li, Y.; Lv, C.; Xue, J. A Novel Predictive Haptic Control Interface for Automation-to-Human Takeover of Automated Vehicles. In Proceedings of the 2019 IEEE Intelligent Vehicles Symposium, Paris, France, 9–12 June 2019. [Google Scholar]
  106. Okada, K.; Sonoda, K.; Wada, T. Transferring From Automated to Manual Driving When Traversing a Curve via Haptic Shared Control. IEEE Trans. Intell. Veh. 2021, 6, 266–275. [Google Scholar] [CrossRef]
  107. de Winter, J.C.F.; Petermeijer, S.M.; Abbink, D.A. Shared control versus traded control in driving: A debate around automation pitfalls. Ergonomics 2023, 66, 1494–1520. [Google Scholar] [CrossRef]
  108. Ma, S.; Zhang, W.; Yang, Z.; Kang, C.; Wu, C.; Chai, C.; Shi, J.; Zeng, Y.; Li, H. Take over Gradually in Conditional Automated Driving: The Effect of Two-stage Warning Systems on Situation Awareness, Driving Stress, Takeover Performance, and Acceptance. Int. J. Hum.–Comput. Interact. 2020, 37, 352–362. [Google Scholar] [CrossRef]
  109. Zhang, W.; Zeng, Y.; Yang, Z.; Kang, C.; Wu, C.; Shi, J.; Ma, S.; Li, H. Optimal Time Intervals in Two-Stage Takeover Warning Systems with Insight Into the Drivers’ Neuroticism Personality. Front. Psychol. 2021, 12, 601536. [Google Scholar] [CrossRef]
Sustainability 16 08345 g001
Figure 1. Conceptual framework for the impact of autonomous vehicle takeover.
Figure 1. Conceptual framework for the impact of autonomous vehicle takeover.
Sustainability 16 08345 g001
Sustainability 16 08345 g002
Figure 2. Schematic diagram of the AV takeover process.
Figure 2. Schematic diagram of the AV takeover process.
Sustainability 16 08345 g002
Sustainability 16 08345 g003
Figure 3. Search strategy flow chart.
Figure 3. Search strategy flow chart.
Sustainability 16 08345 g003
Sustainability 16 08345 g004
Figure 4. Annual publication statistics (note: developed by the authors).
Figure 4. Annual publication statistics (note: developed by the authors).
Sustainability 16 08345 g004
Table 1. Summary of advantages and disadvantages of various types of unimodal interaction.
Table 1. Summary of advantages and disadvantages of various types of unimodal interaction.
Interaction TypesStrengthsCons
Visual (type of icon)Can be visually displayed on the display device [80]Easy to cause driver distraction, may miss TOR [80,86]
Vision (LED surround type)Ability to hold the driver’s distracted attention [84,85]Due to the large impact of ambient light and frequent flashing, it is easy for the driver to ignore the LED prompt [84,85]
Listening (speech class)Reminders clear and clearIn a distracted state, emotional audible cues consume more attention resources [90]
The sense of touchThe speed at which information is transmittedThe content of the information delivered is limited and susceptible to driver habits [96]
Smell (air body)Effective to increase driver alertness [80]It requires professional training to quickly recognize what it represents [98,101]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Guo, H.; Qiu, H.; Zhou, Y.; Deng, Y. Analysis of Influencing Factors of Level 3 Automated Vehicle Takeover: A Literature Review. Sustainability 2024, 16, 8345. https://doi.org/10.3390/su16198345

AMA Style

Guo H, Qiu H, Zhou Y, Deng Y. Analysis of Influencing Factors of Level 3 Automated Vehicle Takeover: A Literature Review. Sustainability. 2024; 16(19):8345. https://doi.org/10.3390/su16198345

Chicago/Turabian Style

Guo, Hanying, Haoyu Qiu, Yongjiang Zhou, and Yuxin Deng. 2024. "Analysis of Influencing Factors of Level 3 Automated Vehicle Takeover: A Literature Review" Sustainability 16, no. 19: 8345. https://doi.org/10.3390/su16198345

APA Style

Guo, H., Qiu, H., Zhou, Y., & Deng, Y. (2024). Analysis of Influencing Factors of Level 3 Automated Vehicle Takeover: A Literature Review. Sustainability, 16(19), 8345. https://doi.org/10.3390/su16198345

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop