Non-driving Tasks Research Articles

Global risk assessment in an autonomous driving context: Impact on both the car and the driver

Highly automated driving (HAD) is a part of the future ways for the intelligent road mobility. In this framework, some studies have shown that drivers' situational awareness decreases when using HAD. In this HAD context, drivers can engage non driving tasks as reading or sleeping. These non-driving tasks lead to increased reaction time in case of hazardous situations or risky events (hardware, sensor, actuator failures, or front obstacle or crashes, or dense traffic congestion, or adverse conditions). In this paper, a global risk indicator is proposed using local information coming from surrounding vehicles or infrastructures (V2X communication). This paper shows firstly the gain of such a global risk indicator comparatively to the local one, and secondly its impact on the behaviour of both the autonomous car and the driver.

The Effects of Lead Time of Take-Over Request and Nondriving Tasks on Taking-Over Control of Automated Vehicles

Automated vehicles have received great attention, since they offer the possibility of significantly increasing traffic safety, mobility, and driver comfort. Current automation technology is still imperfect; therefore, there will still be situations in which the automation will not be able to handle and will request the driver to suspend nondriving tasks and take over control of the automated vehicle in a limited period of time. Accordingly, it is necessary to understand the effects of the lead time of take-over request as well as nondriving tasks on driver take-over. The present driving simulator experiment studied the effects of lead time and various realistic nondriving tasks on take-over behavior and driver acceptance to the automated vehicle. Results suggested optimal driver take-over performance when the lead time of the take-over request was 10–60 s for general nondriving tasks. Specifically, a take-over request with lead time at 10–60 s led to lower crash rate, greater minimum time-to-collision, and lower lateral acceleration. However, a longer lead time (e.g., 15–60 s) was necessary to achieve optimal driver acceptance even though drivers could successfully take over control with shorter lead time (e.g., 10 s). In addition, driver take over performance was significantly influenced by nondriving tasks. When more sensory modalities were occupied or when the cognitive load was very low, driver take-over performance was impaired, especially when the take-over request was too late (e.g., lead time was 3 s). Potential applications of the results in designing of take-over request in automated vehicles were further discussed.

Comparisons of Whole Body Vibration, Muscle Activity and Non-driving Task Performance between Different Seat Suspensions in an Autonomous Passenger Car Application

This study compared whole body vibration (WBV), muscle activity and non-driving task performance between different seat suspension settings in a simulated autonomous passenger car environment. To simulate autonomous vehicle environment, field-measured vibration profiles were recreated on a large-scale 6-degree-of-freedom motion platform. In a repeated-measures laboratory experiment, we measured whole body vibration, muscle activity (neck, shoulder and low back), participants non-driving task performance while participants performed non-driving tasks (pointing task with a laptop trackpad, keyboard typing, web-browsing, and reading) on three different suspension seats mounted on the motion platform: vertical (z-axis) electromagnetic active suspension, multi-axial (lateral (y-axis) and vertical (z-axis)) electromagnetic active suspension, and no suspension (industry standard suspension-less seat for passenger cars). The average weighted vibration [A(8)] and vibration dose value [VDV(8)] showed that the seat measured vibration on both the vertical [A(8) = 0.29 m/s2 and VDV(8) = 10.70 m/s1.75] and multi-axial suspension seats [A(8) = 0.29 m/s2 and VDV(8) = 10.22m/s1.75] were lower than no-suspension seat vibration [A(8) = 0.36 m/s2 and VDV(8) = 12.84 m/s1.75]. Despite the significant differences in WBV between the different suspensions there were no significant differences across three different suspension seats in typing performance (typing speed and accuracy: p’s > 0.83), pointing task performance (movement time and accuracy: p’s > 0.87), web-browsing (number of questions and webpages read: p = 0.42), and reading (number of words read: p = 0.30). The muscle activity in low back (erector spinae) and shoulder (trapezius) muscles also did not show any significant differences (p’s > 0.22). These laboratory study findings indicated that despite the significant reduction in WBV, neither vertical nor multi-axial active suspension seats improve non-driving task performance as compared to the no-suspension seat.

Does automated driving affect time-to-collision judgments?

Automation presumably frees cognitive resources because drivers do not have to control the vehicle. Those resources may be reallocated to processing visual information relevant to driving, such as optic flow, which is relevant for judgments of time-to-collision (TTC). On the other hand, drivers may not use cognitive resources freed during automation to process information relevant to the driving task and improve performance. Drivers may choose to allocate cognitive resources freed during automation to non-driving, secondary tasks (Merat, Jamson, Lai, & Carsten, 2012; Rudin-Brown & Parker, 2004). Therefore, automated driving may lead to performance decrements, particularly when drivers need to resume manual control of the vehicle (Strand, Nilsson, Karlsson, & Nilsson, 2014). The current study compared TTC judgments between automated and manual driving, using a prediction-motion (PM) task which presumably relies on cognitive resources (Tresilian, 1995). We included a braking task to determine whether we could replicate prior reports that drivers brake later during automated driving compared to manual driving (Rudin-Brown & Parker, 2004; de Winter, Happee, Martens & Stanton, 2014). Including PM and braking tasks let us determine whether automation affected only responses (i.e., brake reaction time) or also affected visual perception (i.e., TTC estimation). We hypothesized that automation would affect perceptual judgments rather than solely responses. We expected TTC judgments to be more accurate during automated driving compared to manual driving. We also expected that adding a secondary task that demands cognitive resources would be more detrimental to TTC judgments during automation because the driver would place more cognitive resources on the secondary task during automation than when manually controlling the vehicle. With a driving simulator, participants completed eight drives using manual or automated driving. During half of the drives, participants completed a secondary task, the twenty questions task (TQT), in addition to driving. The TQT is presumably similar to a cell phone conversation because it uses a “question and answer” format (Horrey, Lesch, & Garabet, 2009; Merat et al., 2012, p. 765). At the end of each drive, a critical incident occurred. A vehicle directly in front of the participant’s vehicle decelerated at a rate faster than the automation was capable of braking. Therefore, the automation did not respond to this vehicle’s deceleration. In the braking task, participants used the brake pedal to avoid collision with the lead vehicle. In the PM task, the lead vehicle decelerated for between 0.24 and 3.04 s and then the screen went black. Participants pressed a button to indicate when they thought their vehicle would have hit the lead vehicle if the vehicles’ motions continued in the same manner after the screen went black. Results suggest that automation can affect perceptual judgments in addition to driving responses (e.g., braking). TTC judgments were more accurate, and brake reaction time was faster, during automated driving than manual driving. This occurred even while performing a cognitively-demanding secondary task, suggesting that participants used resources freed by automation to process visual information relevant to TTC judgments rather than complete non-driving tasks. To realize this safety benefit, it is important to design automated systems so that freed cognitive resources are assigned to information relevant to the driving task and not to non-driving tasks.

Interaction between pedestrians and automated vehicles: A Wizard of Oz experiment

Automated vehicles (AVs) will be introduced on public roads in the future, meaning that traditional vehicles and AVs will be sharing the urban space. There is currently little knowledge about the interaction between pedestrians and AVs from the point of view of the pedestrian in a real-life environment. Pedestrians may not know with which type of vehicle they are interacting, potentially leading to stress and altered crossing decisions. For example, pedestrians may show elevated stress and conservative crossing behavior when the AV driver does not make eye contact and performs a non-driving task instead. It is also possible that pedestrians assume that an AV would always yield (leading to short critical gaps). This study aimed to determine pedestrians’ crossing decisions when interacting with an AV as compared to when interacting with a traditional vehicle. We performed a study on a closed road section where participants (N = 24) encountered a Wizard of Oz AV and a traditional vehicle in a within-subject design. In the Wizard of Oz setup, a fake ‘driver’ sat on the driver seat while the vehicle was driven by the passenger by means of a joystick. Twenty scenarios were studied regarding vehicle conditions (traditional vehicle, ‘driver’ reading a newspaper, inattentive driver in a vehicle with “self-driving” sign on the roof, inattentive driver in a vehicle with “self-driving” signs on the hood and door, attentive driver), vehicle behavior (stopping vs. not stopping), and approach direction (left vs. right). Participants experienced each scenario once, in a randomized order. This allowed assessing the behavior of participants when interacting with AVs for the first time (no previous training or experience). Post-experiment interviews showed that about half of the participants thought that the vehicle was (sometimes) driven automatically. Measurements of the participants’ critical gap (i.e., the gap below which the participant will not attempt to begin crossing the street) and self-reported level of stress showed no statistically significant differences between the vehicle conditions. However, results from a post-experiment questionnaire indicated that most participants did perceive differences in vehicle appearance, and reported to have been influenced by these features. Future research could adopt more fine-grained behavioral measures, such as eye tracking, to determine how pedestrians react to AVs. Furthermore, we recommend examining the effectiveness of dynamic AV-to-pedestrian communication, such as artificial lights and gestures.

The effect of peripheral visual feedforward system in enhancing situation awareness and mitigating motion sickness in fully automated driving

This study investigates the impact of peripheral visual information in alleviating motion sickness when engaging in non-driving tasks in fully automated driving. A peripheral visual feedforward system (PVFS) was designed providing information about the upcoming actions of the automated car in the periphery of the occupant’s attention. It was hypothesized that after getting the information from the PVFS, the users’ situation awareness is improved while motion sickness is prevented from developing. The PVFS was also assumed not to increase mental workload nor interrupt the performance of the non-driving tasks. The study was accomplished on an actual road using a Wizard of Oz technique deploying an instrumented car that behaved like a real fully automated car. The test rides using the current setup and methodology indicated high consistency in simulating the automated driving. Results showed that with PVFS, situation awareness was enhanced and motion sickness was lessened while mental workload was unchanged. Participants also indicated high hedonistic user experience with the PVFS. While providing peripheral information showed positive results, further study such as delivering richer information and active head movement are possibly needed.

Asleep at the automated wheel—Sleepiness and fatigue during highly automated driving

Due to the lack of active involvement in the driving situation and due to monotonous driving environments drivers with automation may be prone to become fatigued faster than manual drivers (e.g. Schömig et al., 2015). However, little is known about the progression of fatigue during automated driving and its effects on the ability to take back manual control after a take-over request. In this driving simulator study with Nö=ö60 drivers we used a three factorial 2ö×ö2ö×ö12 mixed design to analyze the progression (12ö×ö5ömin; within subjects) of driver fatigue in drivers with automation compared to manual drivers (between subjects). Driver fatigue was induced as either mainly sleep related or mainly task related fatigue (between subjects). Additionally, we investigated the drivers’ reactions to a take-over request in a critical driving scenario to gain insights into the ability of fatigued drivers to regain manual control and situation awareness after automated driving.Drivers in the automated driving condition exhibited facial indicators of fatigue after 15 to 35ömin of driving. Manual drivers only showed similar indicators of fatigue if they suffered from a lack of sleep and then only after a longer period of driving (approx. 40ömin). Several drivers in the automated condition closed their eyes for extended periods of time. In the driving with automation condition mean automation deactivation times after a take-over request were slower for a certain percentage (about 30%) of the drivers with a lack of sleep (Mö=ö3.2; SDö=ö2.1ös) compared to the reaction times after a long drive (Mö=ö2.4; SDö=ö0.9ös). Drivers with automation also took longer than manual drivers to first glance at the speed display after a take-over request and were more likely to stay behind a braking lead vehicle instead of overtaking it.Drivers are unable to stay alert during extended periods of automated driving without non-driving related tasks. Fatigued drivers could pose a serious hazard in complex take-over situations where situation awareness is required to prepare for threats. Driver fatigue monitoring or controllable distraction through non-driving tasks could be necessary to ensure alertness and availability during highly automated driving.

A Comparative Study on the Detectability between Haptic and Auditory Signals under Driving and Non-driving Task Conditions

A Comparative Study on the Detectability between Haptic and Auditory Signals under Driving and Non-driving Task Conditions

The Effects of Vibration Patterns of Take-Over Request and Non-Driving Tasks on Taking-Over Control of Automated Vehicles

ABSTRACTAutomated vehicles offer the possibility of significantly increasing traffic safety, mobility, and driver comfort, and reducing congestion and fuel emissions. Current automation technology, however, remains imperfect, and in certain situations, automation will still require the driver to suspend non-driving tasks and take back control of the automated vehicle in a limited period of time. During automated driving, drivers engaged in non-driving tasks (e.g., reading, taking a nap) may not perceive the visual or auditory take-over request in a timely nor accurate manner. Therefore, it is necessary to explore the potential of tactile warning further. This study investigates the effects of vibration patterns of take-over requests (six vibration patterns with different orders of the vibration location) and various realistic non-driving tasks (six non-driving tasks: reading, typing, watching videos, playing games, taking a nap, and monitoring the driving scenario on the driving simulator) on driver take-over behavior, and driver trust and acceptance of automated vehicles. Across all non-driving tasks, the fastest response time was observed with Vibration Pattern 5 (order of the vibration location: back–back–seat–seat). The shortest response time and largest minimum time-to-collision (TTC) also were observed when drivers took back control of the vehicle after monitoring the driving scenario. No interaction effects between vibration patterns and non-driving tasks were observed. Potential applications of the results of designing take-over requests in automated vehicles are discussed.

How safe is tuning a radio?: using the radio tuning task as a benchmark for distracted driving

Drivers engage in non-driving tasks while driving, such as interactions entertainment systems. Studies have identified glance patterns related to such interactions, and manual radio tuning has been used as a reference task to set an upper bound on the acceptable demand of interactions. Consequently, some view the risk associated with radio tuning as defining the upper limit of glance measures associated with visual-manual in-vehicle activities. However, we have little knowledge about the actual degree of crash risk that radio tuning poses and, by extension, the risk of tasks that have similar glance patterns as the radio tuning task. In the current study, we use counterfactual simulation to take the glance patterns for manual radio tuning tasks from an on-road experiment and apply these patterns to lead-vehicle events observed in naturalistic driving studies. We then quantify how often the glance patterns from radio tuning are associated with rear-end crashes, compared to driving only situations. We used the pre-crash kinematics from 34 crash events from the SHRP2 naturalistic driving study to investigate the effect of radio tuning in crash-imminent situations, and we also investigated the effect of radio tuning on 2,475 routine braking events from the Safety Pilot project. The counterfactual simulation showed that off-road glances transform some near-crashes that could have been avoided into crashes, and glance patterns observed in on-road radio tuning experiment produced 2.85–5.00 times more crashes than baseline driving.

Assessing drivers' response during automated driver support system failures with non-driving tasks

IntroductionWith the increase in automated driver support systems, drivers are shifting from operating their vehicles to supervising their automation. As a result, it is important to understand how drivers interact with these automated systems and evaluate their effect on driver responses to safety critical events. This study aimed to identify how drivers responded when experiencing a safety critical event in automated vehicles while also engaged in non-driving tasks. MethodIn total 48 participants were included in this driving simulator study with two levels of automated driving: (a) driving with no automation and (b) driving with adaptive cruise control (ACC) and lane keeping (LK) systems engaged; and also two levels of a non-driving task (a) watching a movie or (b) no non-driving task. In addition to driving performance measures, non-driving task performance and the mean glance duration for the non-driving task were compared between the two levels of automated driving. ResultsDrivers using the automated systems responded worse than those manually driving in terms of reaction time, lane departure duration, and maximum steering wheel angle to an induced lane departure event. These results also found that non-driving tasks further impaired driver responses to a safety critical event in the automated system condition. ConclusionIn the automated driving condition, driver responses to the safety critical events were slower, especially when engaged in a non-driving task. Practical applicationTraditional driver performance variables may not necessarily effectively and accurately evaluate driver responses to events when supervising autonomous vehicle systems. Thus, it is important to develop and use appropriate variables to quantify drivers' performance under these conditions.

Changes over 12 months in eye glances during secondary task engagement among novice drivers

During their first year of driving, crash rates among novice drivers are very high but decline rapidly. However, it is not clear what skills or knowledge they are acquiring in this period. Secondary task engagement while driving is a contributing factor to many traffic collisions and some of the elevated crash risk among novices could be explained by greater prevalence or longer periods of eyes off the road while engaging in these non-driving tasks. The current study looked at the eye glances of novice teen drivers engaging in secondary tasks on a test track at 0 and 12 months of licensure and compared their performance with their parents. Novices improved from 0 to 12 months on their longest single glance off the forward roadway and total percentage of time for eyes off the forward roadway, but parents remained stable. Compared with their parents, the longest single glance off the forward roadway was longer for novices at 0 months, but by 12 months there was no difference between the groups. However, for total percentage of time for eyes off the forward roadway, novices performed the same as their parents at 0 months and actually had shorter times at 12 months. These findings could reflect the combined development of driving skills over 12 months and the relative experience that modern teenagers have with portable electronic devices. The results suggest that novice drivers are particularly poor at engaging with secondary tasks while driving.

Vibrotactile Displays: A Survey With a View on Highly Automated Driving

The task of car driving is automated to an ever greater extent. In the foreseeable future, drivers will no longer be required to touch the steering wheel and pedals and could engage in non-driving tasks such as working or resting. Vibrotactile displays have the potential to grab the attention of the driver when the automation reaches its functional limits and the driver has to take over control. The aim of the present literature survey is to outline the key physiological and psychophysical aspects of vibrotactile sensation and to provide recommendations and relevant research questions regarding the use of vibrotactile displays for taking over control from an automated vehicle. Results showed that a distinction can be made between four dimensions for coding vibrotactile information (amplitude, frequency, timing, and location), each of which can be static or dynamic. There is a consensus that frequency and amplitude are less suitable for coding information than location and timing. Vibrotactile stimuli have been shown to be effective as simple warnings. However, vibrations can evoke annoyance, and providing vibrations in close spatial–temporal proximity might cause a lack of comprehension of the signal. We describe the sequential stages of a take-over process and argue that vibrotactile displays are a promising candidate for redirecting the attention of a distracted driver. Furthermore, vibrotactile displays hold potential for supporting cognitive processing and action selection while resuming control of an automated vehicle. Finally, we argue that multimodal feedback should be used to assist the driver in the take-over process.

The influence of age on the take-over of vehicle control in highly automated driving

The growing proportion of older drivers in the population plays an increasingly relevant role in road traffic that is currently awaiting the introduction of automated vehicles. In this study, it was investigated how older drivers (⩾60years) compared to younger drivers (⩽28years) perform in a critical traffic event when driving highly automated. Conditions of the take-over situation were manipulated by adding a verbal non-driving task (20 questions task) and by variation of traffic density. Two age groups consisting of 36 younger and 36 older drivers drove either with or without a non-driving task on a six-lane highway. They encountered three situations with either no, medium or high traffic density where they had to regain vehicle control and evade an obstacle on the road. Older drivers reacted as fast as younger drivers, however, they differed in their modus operandi as they braked more often and more strongly and maintained a higher time-to-collision (TTC). Deterioration of take-over time and quality caused by increased traffic density and engagement in a non-driving task was on the same level for both age groups. Independent of the traffic density, there was a learning effect for both younger and older drivers in a way that the take-over time decreased, minimum TTC increased and maximum lateral acceleration decreased between the first and the last situation of the experiment. Results highlight that older drivers are able to solve critical traffic events as well as younger drivers, yet their modus operandi differs. Nevertheless, both age groups adapt to the experience of take-over situations in the same way.

A Study of a Driver Model Using Unconscious Driving Behaviors

The automation of automobiles will change driver behavior, as the unconscious region stressed by the conscious region. Driving operation is considered to be done unconsciously, as human biological signals start 0.3seconds before a decision is made. It is thought that driving operation changes when the unconscious region is pressured. In this paper, we set mental calculations as a dual task technique of a non-driving task for drivers, and verified the changing of driver operations by this load pressure on a driver's consciousness region.

Modeling one-choice and two-choice driving tasks.

An experiment is presented in which subjects were tested on both one-choice and two-choice driving tasks and on non-driving versions of them. Diffusion models for one- and two-choice tasks were successful in extracting model-based measures from the response time and accuracy data. These include measures of the quality of the information from the stimuli that drove the decision process (drift rate in the model), the time taken up by processes outside the decision process and, for the two-choice model, the speed/accuracy decision criteria that subjects set. Drift rates were only marginally different between the driving and non-driving tasks, indicating that nearly the same information was used in the two kinds of tasks. The tasks differed in the time taken up by other processes, reflecting the difference between them in response processing demands. Drift rates were significantly correlated across the two two-choice tasks showing that subjects that performed well on one task also performed well on the other task. Nondecision times were correlated across the two driving tasks, showing common abilities on motor processes across the two tasks. These results show the feasibility of using diffusion modeling to examine decision making in driving and so provide for a theoretical examination of factors that might impair driving, such as extreme aging, distraction, sleep deprivation, and so on.

Driver State Monitoring Systems– Transferable Knowledge Manual Driving to HAD

Highly Automated Driving (HAD) will be commercially available in a near feature, yet human factors issues like the influence of driver state can have a critical impact in the success of this driving paradigm and also in road safety. It is very likely that Driver State Monitoring Systems (DSMS) will play a bigger role than they have played so far. However with this new driving paradigm shift is important to select highlight what is transferable from the previous systems. Due to lack of driving task engagement, driving performance metrics are no longer viable, creating opportunities for other approaches like detecting non-driving task engagement or fatigue countering behaviours. Eye based metrics will remain important.

Effects of adaptive cruise control and highly automated driving on workload and situation awareness: A review of the empirical evidence

Adaptive cruise control (ACC), a driver assistance system that controls longitudinal motion, has been introduced in consumer cars in 1995. A next milestone is highly automated driving (HAD), a system that automates both longitudinal and lateral motion. We investigated the effects of ACC and HAD on drivers’ workload and situation awareness through a meta-analysis and narrative review of simulator and on-road studies. Based on a total of 32 studies, the unweighted mean self-reported workload was 43.5% for manual driving, 38.6% for ACC driving, and 22.7% for HAD (0%=minimum, 100=maximum on the NASA Task Load Index or Rating Scale Mental Effort). Based on 12 studies, the number of tasks completed on an in-vehicle display relative to manual driving (100%) was 112% for ACC and 261% for HAD. Drivers of a highly automated car, and to a lesser extent ACC drivers, are likely to pick up tasks that are unrelated to driving. Both ACC and HAD can result in improved situation awareness compared to manual driving if drivers are motivated or instructed to detect objects in the environment. However, if drivers are engaged in non-driving tasks, situation awareness deteriorates for ACC and HAD compared to manual driving. The results of this review are consistent with the hypothesis that, from a Human Factors perspective, HAD is markedly different from ACC driving, because the driver of a highly automated car has the possibility, for better or worse, to divert attention to secondary tasks, whereas an ACC driver still has to attend to the roadway.

Task analysis of transit bus drivers’ left-turn maneuver: Potential countermeasures for the reduction of collisions with pedestrians

Left-turn maneuvers by transit buses are associated with a particularly high rate of collisions with pedestrians. The primary objective of the current study was, through task analysis, to gather a detailed outline of all mental and perceptual tasks that transit bus drivers complete when making a left turn at a signalized intersection. As a secondary objective, based on the task analysis examination, we proposed several countermeasures for the reduction of collision between transit buses and pedestrians. The task analysis revealed that transit bus drivers engage in various tasks during all stages of a left-turn maneuver. The results also showed that the task demand is at a peak during the performance of one of the most critical tasks – detecting pedestrians. Non-driving tasks that transit bus drivers are required to perform (e.g., collecting fares, providing directions to passengers) further increase the tasks demands. Based on the results of the task analysis, we proposed several potential solutions for the reduction of collisions between transit buses and pedestrians. The proposed countermeasures included a technology-based (collision warning system for both drivers and the pedestrians) and infrastructure-based (protected left turn signal for buses only) solutions, as well as examination of individual factors (e.g., stress coping ability) and the relationship with different aspects of driving performance.

Control Task Substitution in Semiautomated Driving

The study was designed to show how driver attention to the road scene and engagement of a choice of secondary tasks are affected by the level of automation provided to assist or take over the basic task of vehicle control. It was also designed to investigate the difference between support in longitudinal control and support in lateral control. There is comparatively little literature on the implications of automation for drivers' engagement in the driving task and for their willingness to engage in non-driving-related activities. A study was carried out on a high-level driving simulator in which drivers experienced three levels of automation: manual driving, semiautomated driving with either longitudinal or lateral control provided, and highly automated driving with both longitudinal and lateral control provided. Drivers were free to pay attention to the roadway and traffic or to engage in a range of entertainment and grooming tasks. Engagement in the nondriving tasks increased from manual to semiautomated driving and increased further with highly automated driving. There were substantial differences in attention to the road and traffic between the two types of semiautomated driving. The literature on automation and the various task analyses of driving do not currently help to explain the effects that were found. Lateral support and longitudinal support may be the same in terms of levels of automation but appear to be regarded rather differently by drivers.