Publication Hub Archive

Driving Simulation

You have reached the Ergoneers Publication Hub for:

Field of Application > Driving Simulation

Find all Publications here:

Publication Hub

Total results: 274

Classification of driver cognitive load based on physiological data: Exploring recurrent neural networks

Year: 2022

Authors: S Kumar,D He, G Qiao

In-vehicle systems can lead to high cognitive load that impairs driving performance. Interfaces that can detect and adapt to cognitive load accordingly may alleviate these effects. Previous research explored machine learning models to classify drivers’ cognitive load based on physiological signals but most conducted training and testing on data from the same participants (i.e., within-driver partitioning), which raises generalizability and practical feasibility concerns. In this paper, we explored the performance of widely-used models by training and testing them on data from different subjects (i.e., across-drivers partitioning), and further compared them with a more recent model that is effective for time-series data, the recurrent neural network (RNN). A driving simulator dataset was used to classify 2 levels of cognitive load (external cognitive secondary task vs. no task). All models performed better with within-driver partitioning. RNN outperformed other models with mean accuracies of 88.1% and 85.6% with within-driver and across-drivers partitioning, respectively.

4 versions available

Classification of driver cognitive load: Exploring the benefits of fusing eye-tracking and physiological measures

Year: 2022

Authors: D He, Z Wang,EB Khalil,B Donmez

In-vehicle infotainment systems can increase cognitive load and impair driving performance. These effects can be alleviated through interfaces that can assess cognitive load and adapt accordingly. Eye-tracking and physiological measures that are sensitive to cognitive load, such as pupil diameter, gaze dispersion, heart rate (HR), and galvanic skin response (GSR), can enable cognitive load estimation. The advancement in cost-effective and nonintrusive sensors in wearable devices provides an opportunity to enhance driver state detection by fusing eye-tracking and physiological measures. As a preliminary investigation of the added benefits of utilizing physiological data along with eye-tracking data in driver cognitive load detection, this paper explores the performance of several machine learning models in classifying three levels of cognitive load imposed on 33 drivers in a driving simulator study: no external load, lower difficulty 1-back task, and higher difficulty 2-back task. We built five machine learning models, including k-nearest neighbor, support vector machine, feedforward neural network, recurrent neural network, and random forest (RF) on (1) eye-tracking data only, (2) HR and GSR, (3) eye-tracking and HR, (4) eye-tracking and GSR, and (5) eye-tracking, HR, and GSR. Although physiological data provided 1%–15% lower classification accuracies compared with eye-tracking data, adding physiological data to eye-tracking data increased model accuracies, with an RF classifier achieving 97.8% accuracy. GSR led to a larger boost in accuracy (29.3%) over HR (17.9%), with the combination of the two factors boosting accuracy by 34.5%. Overall, utilizing both physiological and eye-tracking measures shows promise for driver state detection applications.

7 versions available

Does a faster takeover necessarily mean it is better? A study on the influence of urgency and takeover-request lead time on takeover performance and safety

Year: 2022

Authors: H Wu, C Wu, N Lyu, J Li

During conditionally automated driving, drivers are sometimes required to take over control of the vehicle if a so-called takeover request (TOR) is issued. TORs are generally issued due to system limitations. This study investigated the effect of different urgency scenarios and takeover-request lead times (TORlts) on takeover performance and safety. The experiment was conducted in a real vehicle-based driving simulator. Manual driving, 7-second TORlt and 5-second TORlt were each tested. Participants experienced three progressively urgent driving scenarios: one cut-in scenario and two obstacle-avoidance scenarios. The results indicate that the TORlt significantly affected takeover performance and safety. Within a certain range, the longer the TORlt, the safer the takeover. However, while takeover reaction time depended mainly on the length of the TORlt and was not significantly related to other factors, such as workload, greater workloads that were caused by the TORlt were associated with shorter reaction times and decreased safety. This is evidence that the reaction time should not be used as the preferred indicator to evaluate takeover performance and safety. Indicators, such as workload, minimum TTC, feature point distribution position and slope of the obstacle avoidance trajectory, can better measure and evaluate takeover performance and safety. This study can provide data support for takeover safety evaluation of conditionally automated driving.

4 versions available

Does gender affect the driving performance of young patients with diabetes?

Year: 2022

Authors: s

Recent evidence suggests that poor glycemic control among young patients with type 1 diabetes mellitus (T1DM) has negative cognitive and physical effects, whose extent is gender-dependent. For example, female patients with diabetes present more physical and cognitive limitations than male patients in terms of cognitive adjustment, quality of decision making, and functioning. Studies about traffic safety report that diabetic drivers are at increased risk of being involved in road crashes, especially when driving in a state of hypoglycemia under which their blood glucose level is too low. We have recently demonstrated that acute hyperglycemia (when the blood glucose level is too high) can also lead to poor driving performance among T1DM young adult patients. Against this background, the objective of the present study was to find out whether gender affects the driving performance of young drivers with diabetes. Twenty-six T1DM drivers participated in a counterbalanced crossover experiment. While being monitored by an eye tracker, they drove a driving simulator and twice navigated through the nine hazardous scenarios: once under a normal blood glucose (euglycemia) level and once high blood glucose (hyperglycemia) level. The first main result is that young female drivers are more affected by diabetes than young male drivers, regardless of momentary glycemic changes. The second main result is that poor glycemic control substantially deteriorates hazard perception and driving performance of young males with diabetes. Thus, it is argued that an uncontrolled state of a high blood glucose level may be more hazardous for young males with diabetes since it negatively impacts their driving performance.

7 versions available

Evaluation of a dynamic blocking concept to mitigate driver distraction: three simulator studies

Year: 2022

Authors: J Leipnitz, A Gross, J Dostert, T Baumgarten

In recent years, the number and complexity of in-vehicle infotainment systems has been steadily increasing. While these systems certainly improve the driving experience, they also increase the risk for driver distraction. International standards and guidelines provide methods of measuring this distraction along with test criteria that help automakers decide whether an interface task is too distracting to be used while driving. Any specific function failing this test should therefore be locked out for use by the driver. This study implemented and tested a dynamic approach to this blocking by algorithmically reacting to driver inputs and the pace of the interaction in order to prevent drivers from having prolonged or too intense sequences of in-vehicle interactions not directly related to driving. Three simulated driving experiments in Germany and the United States were conducted to evaluate this dynamic function blocking concept and also cater for differences in the status quo of either no blocking or static blocking. The experiments consisted of a car following scenario with various secondary interface tasks and always included a baseline condition where no blocking occurred as well as an implementation of the dynamic function blocking. While Experiments 1 and 3 were aimed at collecting and analyzing gaze and driving data from more than 20 participants, Experiment 2 focused on the user experience evaluation of different visual feedback implementations from 13 participants. The user experience as rated by these participants increased throughout the course of all three studies and helped further improve both the concept and feedback design. In the experiments the total glance time towards the road was significantly higher in the dynamic function blocking condition compared to the baseline, already accounting for the increase in total task time inherent to the dynamic condition. Participants developed two strategies of interacting with the dynamic function blocking. They either operated at their normal baseline speed and incurred task blockings or operated slower to avoid the blockings. In the latter strategy, participants chunked their interactions into smaller steps with the present data suggesting that they used the pauses in between chunks to look back onto the road ahead. Theoretical and practical implications of this first evaluation of a dynamic function blocking concept are discussed.

6 versions available

Evaluation of the optimal quantity of in-vehicle information icons using a fuzzy synthetic evaluation model in a driving simulator

Year: 2022

Authors: J Chen, X Wang, Z Cheng, Y Gao

In-Vehicle Information (IVI) features such as navigation assistance play an important role in the travel of drivers around the world. Frequent use of IVI, however, can easily increase the cognitive load of drivers. The interface design, especially the quantity of icons presented to the driver such as those for navigation, music, and phone calls, has not been fully researched. To determine the optimal number of icons, a systematic evaluation of the IVI Human Machine Interface (HMI) was examined using single-factor and multivariate analytical methods in a driving simulator. When one-way ANOVA was performed, the results showed that the 3-icon design scored best in subjective driver assessment, and the 4-icon design was best in the steering wheel angle. However, when a new method of analyzing the data that enabled a simultaneous accounting of changes observed in the dependent measures, 3 icons had the highest score (that is, revealed the overall best performance). This method is referred to as the fuzzy synthetic evaluation model (FSE). It represents the first use of it in an assessment of the HMI design of IVI. The findings also suggest that FSE will be applicable to various other HMI design problems.

4 versions available

Eye tracking system measurement of saccadic eye movement with different illuminance transmission exposures during driving simulation

Year: 2022

Authors: A Ahmad,SA Rosli,AH Chen

Numerous eye gaze changes of different fixation viewings are involved in driving. In addition, driving is done under various surrounding illuminance conditions. However, the effect of different illuminance transmissions on eye gaze movement was not explored during driving. This study investigated the saccadic eye movement using eye tracking system under different illuminance transmissions during driving simulation. The investigation was conducted on twenty-eight participants aged between 21 to 26 years old with proper driving licensing and experience. All participants had good vision status, with a good history of systemic, ocular, and binocular vision health. Using driving simulation, the participants were instructed to drive as they usually did, and their saccadic eye movement was recorded via the Dikablis eye tracker. The surrounding illuminance within the experimental room provided 100% transmission of 500 Lux, and the illuminance transmission was varied to 50%, 30%, and 15% using neural density filters. Under different illuminance transmissions, the saccadic eye movement showed no significant differences (p>0.05), even with the 15% transmission, both in the number and duration of saccadic eye movement. This showed similar eye gaze change specifically saccadic movement during driving simulation with different light transmissions. It could be concluded that eye gaze movement was not influenced by reduced illuminance when driving.

1 version available:

How does navigating with Augmented Reality information affect drivers’ glance behaviour in terms of attention allocation?

Year: 2022

Authors: K Bauerfeind, J Drüke, L Bendewald

Drivers can benefit from Augmented Reality (AR) information especially in ambiguous navigation situations compared to conventional Head-up displays (HUD). AR information is correctly superimposed on the relevant objects in the environment and therefore directly related to the driving situation. Hence, it is assumed, that drivers no longer have to switch glances between the AR information and the environment (Kim & Dey, 2009). It has to be investigated whether switching glances between the presented navigation information and the environment can be reduced with AR information compared to HUD information. Furthermore, the question arises whether AR information might capture drivers’ attention and therefore distract from the traffic situation compared to a HUD as AR information is presented on the driver’s primary visual axis. The aim of the driving simulator study was to examine glance behaviour in terms of attention allocation while participants navigated in an ambiguous left turn situation with an oncoming car in an urban area (N = 58). Hence, drivers were faced with the decision to turn in front of it or let it pass. A conventional HUD and an AR display presented the navigation information to the driver. The drives differed in traffic complexity (low vs. high) to provide indications whether drivers adapt glance behaviour to altered environmental conditions. Besides the navigation task, drivers performed a non-driving-related task to raise drivers’ mental load while navigating. Results showed that with the AR display participants paid more attention to an oncoming car in the ambiguous left turn situation than with the HUD, which indicates that AR information was not distracting. Furthermore, participants switched glances significantly less between the AR navigation information and the environment, which indicates that with the AR display the driver did not have to map the virtual information onto the real driving situation. Independently of the display type 88% of the participants let the oncoming car pass the first time in this situation. Moreover, subjective data showed that drivers benefitted from AR information. The results of this study contribute to the investigation and development of AR user interfaces.

2 versions available

Interaction strategies with advanced driver assistance systems

Year: 2022

Authors: N Neuhuber,P Pretto,B Kubicek

When using advanced driver assistance systems (ADAS) drivers need to calibrate their level of trust and interaction strategy to changes in the driving context and possible consequent reduction of system reliability (e.g. in harsh weather conditions). By investigating and identifying categories of drivers who choose inadequate interaction strategies, it is possible to address unsafe usage with e.g. tutoring lessons tailored to the respective driver category. This paper presents two studies investigating categories of drivers who apply different interaction strategies when using ADAS. Study I was designed as an exploratory field study with 37 participants interacting with a SAE level 2 system. For the exploratory study, it was important to observe and understand the interaction strategies in a driving context which entails the real complexity of the driving task. The experimental set-up of study II (simulator study), however, allowed to clearly interpret the interaction strategies as either calibrated or un-calibrated by varying the situational risk. Participants (N = 33) were driving in a situation where the system was either working reliably (low-risk condition) or in a situation where the system displayed repeatedly errors under harsh weather conditions (high-risk condition). Cluster analyses with the variables trust, monitoring behavior towards the system and usage behavior were performed to analyze potential categories of drivers. Extreme driver categories with interaction strategies indicative for both misuse and disuse were observed in both studies. In study I, drivers were categorized as either highly trusting attentive, moderately trusting attentive, moderately inattentive, inattentive or skeptical. In study II, drivers were categorized as either un-calibrated, calibrated, inconsistent or skeptical. Taken together, results underline the need of tutoring systems that are tailored for different driver categories.

5 versions available

Interruption management in the context of take-over-requests in conditional driving automation

Year: 2022

Authors: A Borowsky,N Zangi

Drivers of partially automated vehicles are relieved from parts of the driving tasks allocated to the automated driver. This reduction in driving demands encourages them to engage with nondriving related tasks, which may impair awareness of the road environment once a takeover request (TOR) is initiated. This article examined the four suggested strategies drivers that take to regain control following a TOR, from the perspective of interruption management principles. Thirty students participated in a simulated study of two drives, where we manipulated TOR alerts, time to regain control, and potential road hazards. We hypothesized that all four interruption management strategies will be observed. Our hypothesis was confirmed. Four strategies were identified. Most drivers chose strategy 2 to accept and initiate the takeover immediately after the TOR started. The second frequent strategy was to reject the TOR but look at the road. Drivers’ strategy choices changed following alert type and the chronological drive order. With simulated driving experience (i.e., second drive), drivers postponed taking control, adapting to the time budget. Yet, inaccurate understanding of the situation or over-trust affected the chosen strategy. We conclude that interruption management principles are beneficial for studying how drivers respond to TORs and evaluating options to improve TOR performance.

2 versions available