Silent failure detection in partial automation as a function of visual attentiveness.

OBJECTIVE: Drivers using level 2 automation are able to disengage with the dynamic driving task, but must still monitor the roadway and environment and be ready to takeover on short notice. However, people are still willing to engage with non-driving related tasks, and the ways in which people manage this tradeoff are expected to vary depending on the operational design domain of the system and the nature of the task. Our aim is to model driver gaze behavior in level 2 partial driving automation when the driver is engaged in an email task on a cell phone. Both congested highway driving, traffic jams, and a hazard with a silent automation failure are considered in a driving simulator study conducted in the NADS-1 high-fidelity motion-based driving simulator. METHODS: Sequence analysis is a methodology that has grown up around social science research questions. It has developed into a powerful tool that supports intuitive visualizations, clustering analysis, covariate analyses, and Hidden Markov Models. These methods were used to create models for four different gaze behaviors and use the models to predict attention during the silent failure event. RESULTS: Predictive simulations were run with initial conditions that matched driver state just prior to the silent failure event. Actual gaze response times were observed to fall within distributions of predicted glances to the front. The three drivers with the largest glance response times were not able to take back manual control before colliding with the hazard. CONCLUSIONS: The simulated glance response time distributions can be used in more sophisticated ways when combined with other data. The glance response time probability may be conditioned on other variables like time on task, time of day, prevalence of the current behavior for this driver, or other variables. Given the flexibility of sequence analysis and the methods it supports (clustering, HMMs), future studies may benefit from its application to gaze behavior and driving performance data.

Autonomous Shuttle Operating on Highways and Gravel Roads in Rural America: A Demonstration Study.

The safe integration of Automated Driving Systems (ADS) into the nation's on-road transportation system, particularly in rural areas, could vastly improve overall quality of life for a rapidly growing segment of the US population. This paper describes findings from the first half (i.e., three of six phases) of a demonstration project called "ADS for Rural America". The goal of this project is to conduct a series of demonstrations that utilizes an autonomous shuttle to show how older adults (≥65 years old) could be transported from their rural homes to other locations in rural areas, as well as an urban center. This paper examines older adults' perceptions of automation before and after riding in an autonomous shuttle and their ratings of anxiety throughout the ride as they experience particular road types and maneuvers. After riding in the shuttle, older adults expressed decreased suspicion, increased trust, and increased reliability of ADS compared to baseline. Older adults reported low levels of anxiety during the 90 min ride in the shuttle. To promote the adoption and acceptance of ADS, older adults should be exposed to this technology.

The Effect of Partial Automation on Driver Attention: A Naturalistic Driving Study.

OBJECTIVE: This naturalistic driving study investigated how drivers deploy visual attention in a partially automated vehicle. BACKGROUND: Vehicle automation is rapidly increasing across vehicle fleets. This increase in automation will likely have both positive and negative consequences as drivers learn to use the new technology. Research is needed to understand how drivers interact with partially automated vehicle systems and what impact new technology has on driver attention. METHOD: Ten participants drove a Tesla Model S for 1 week during their daily commute on a stretch of busy interstate. Drivers were instructed to use Autopilot, a system that provides both lateral and longitudinal control, as much as they felt comfortable while driving on the interstate. Driver-facing video data were recorded and manually reduced to examine glance behavior. RESULTS: Drivers primarily allocated their visual attention between the forward roadway (74% of glance time) and the instrument panel (13%). With partial automation engaged, drivers made longer single glances and had longer maximum total-eyes-off-road time (TEORT) associated with a glance cluster. CONCLUSION: These results provide a window into the nature of visual attention while driving with partial vehicle automation. The results suggest that drivers may be more willing to execute long, "outlier" glances and clusters of glances to off-road locations with partial automation. The findings highlight several important human factors considerations for partially automated vehicles.

Examining teen driver crashes and the prevalence of distraction: Recent trends, 2007-2015.

INTRODUCTION: Teen drivers crash at a much higher rate than adult drivers, with distractions found as a factor in nearly 6 out of 10 moderate-to-severe teen crashes. As the driving environment continues to rapidly evolve, it is important to examine the effect these changes may be having on our youngest and most vulnerable drivers. METHOD: The purpose of this study was to identify types of vehicle crashes teens are most frequently involved in, as well as the distracting activities being engaged in leading up to these crashes, with a focus on identifying changes or trends over time. We examined 2,229 naturalistic driving videos involving drivers ages 16-19. These videos captured crashes occurring between 2007 and 2015. The data of interest for this study included crash type, behaviors drivers engaged in leading up to the collision, total duration of time the driver's eyes were off the forward roadway, and duration of the longest glance away from forward. RESULTS: Rear-end crashes increased significantly (annual % change=3.23 [2.40-4.05]), corresponding with national data trends. Among cell phone related crashes, a significant shift occurred, from talking/listening to operating/looking (annual % change=4.22 [1.15-7.29]). Among rear-end crashes, there was an increase in the time drivers' eyes were off the road (ß=0.1527, P=0.0004) and durations of longest glances away (ß=0.1020, P=0.0014). CONCLUSIONS: Findings suggest that shifts in the way cell phones are being used, from talking/listening to operating/looking, may be a cause of the increasing number of rear-end crashes for teen drivers. PRACTICAL APPLICATIONS: Understanding the role that cell phone use plays in teen driver crashes is extremely important. Knowing how and when teens are engaging in this behavior is the only way effective technologies can be developed for mitigating these crashes.

Using event-triggered naturalistic data to examine the prevalence of teen driver distractions in rear-end crashes.

INTRODUCTION: While teen driver distraction is cited as a leading cause of crashes, especially rear-end crashes, little information is available regarding its true prevalence. The majority of distraction studies rely on data derived from police reports, which provide limited information regarding driver distraction. METHOD: This study examined over 400 teen driver rear-end crashes captured by in-vehicle event recorders. A secondary data analysis was conducted, paying specific attention to driver behaviors, eyes-off-road time, and response times to lead-vehicle braking. RESULTS: Among teens in moderate to severe rear-end crashes, over 75% of drivers were observed engaging in a potentially distracting behavior. The most frequently seen driver behaviors were cell phone use, attending to a location outside the vehicle, and attending to passengers. Drivers using a cell phone had a significantly longer response time than drivers not engaged in any behaviors, while those attending to passengers did not. Additionally, in about 50% of the rear-end crashes where the driver was operating/looking at a phone (e.g., texting), the driver showed no driver response (i.e., braking or steering input) before impact, compared to 10% of crashes where the driver was attending to a passenger. CONCLUSIONS: The high frequency of attending to passengers and use of a cell phone leading up to a crash, compounded with the associated risks, underlines the importance of continued investigation in these areas. PRACTICAL APPLICATIONS: Parents and teens must be educated regarding the frequency of and the potential effects of distractions. Additional enforcement may be necessary if Graduated Driver Licensing (GDL) programs are to be effective. Systems that alert distracted teens could also be especially helpful in reducing rear-end collisions.

Analysis of naturalistic driving videos of fleet services drivers to estimate driver error and potentially distracting behaviors as risk factors for rear-end versus angle crashes.

OBJECTIVE: The objective of this study was to estimate the prevalence and odds of fleet driver errors and potentially distracting behaviors just prior to rear-end versus angle crashes. METHODS: Analysis of naturalistic driving videos among fleet services drivers for errors and potentially distracting behaviors occurring in the 6 s before crash impact. Categorical variables were examined using the Pearson's chi-square test, and continuous variables, such as eyes-off-road time, were compared using the Student's t-test. Multivariable logistic regression was used to estimate the odds of a driver error or potentially distracting behavior being present in the seconds before rear-end versus angle crashes. RESULTS: Of the 229 crashes analyzed, 101 (44%) were rear-end and 128 (56%) were angle crashes. Driver age, gender, and presence of passengers did not differ significantly by crash type. Over 95% of rear-end crashes involved inadequate surveillance compared to only 52% of angle crashes (P < .0001). Almost 65% of rear-end crashes involved a potentially distracting driver behavior, whereas less than 40% of angle crashes involved these behaviors (P < .01). On average, drivers spent 4.4 s with their eyes off the road while operating or manipulating their cell phone. Drivers in rear-end crashes were at 3.06 (95% confidence interval [CI], 1.73-5.44) times adjusted higher odds of being potentially distracted than those in angle crashes. CONCLUSIONS: Fleet driver driving errors and potentially distracting behaviors are frequent. This analysis provides data to inform safe driving interventions for fleet services drivers. Further research is needed in effective interventions to reduce the likelihood of drivers' distracting behaviors and errors that may potentially reducing crashes.

Evaluation of driving behavior on rural 2-lane curves using the SHRP 2 naturalistic driving study data.

INTRODUCTION: Over half of motor vehicle fatalities are roadway departures, with rural horizontal curves being of particular interest because they make up only a small share of the system mileage but have a crash rate that is significantly higher than tangent sections. However the interaction between the driver and roadway environment is not well understood, and, as a result, it is difficult to select appropriate countermeasures. METHOD: In order to address this knowledge gap, data from the SHRP 2 naturalistic driving study were used to develop relationships between driver, roadway, and environmental characteristics and risk of a road departure on rural curves. The SHRP 2 NDS collected data from over 3,000 male and female volunteer passenger vehicle drivers, ages 16-98, during a three year period, with most drivers participating between one to two years. A Roadway Information Database was collected in parallel and contains detailed roadway data collected on more than 12,500 centerline miles of highways in and around the study sites. RESULTS: Roadway data were reduced for rural 2-lane curves and included factors such as geometry, shoulder type, presence of rumble strips, etc. Environmental and traffic characteristics, such as time of day, ambient conditions, or whether the subject vehicle was following another vehicle, were reduced from the forward roadway video view. Driver characteristics, such as glance location and distraction were reduced from the driver and over the shoulder videos. CONCLUSIONS: Logistic regression models were developed to assess the probability (odds) of a given type of encroachment based on driver, roadway, and environmental characteristics. At the point this study was undertaken, crashes and near crashes were not yet available and only around 1/3 of the full SHRP NDS dataset could be queried. As a result, the likelihood of crossing the right or left lane line (encroachments) and speeding were used as dependent variables.

Using an event-triggered video intervention system to expand the supervised learning of newly licensed adolescent drivers.

OBJECTIVES: We examined whether feedback from an event-triggered video intervention system reduced the number of safety-relevant driving errors made by newly licensed adolescents. METHODS: We used a 1-group pretest-posttest quasi-experimental design to compare the rate of coachable error events per 1000 miles for 18 drivers who were aged 16 years. The intervention consisted of immediate visual feedback provided to the drivers and weekly event reports and videos provided to the drivers and their parents. RESULTS: The number of coachable events was reduced by 61% overall during the intervention (chi(2) = 11.42; P = .001) and did not significantly increase during the second baseline, which was assessed after the intervention ended (chi(2) = 1.49; P = .223). The greatest reduction was seen in the category of improper turns or curves and for drivers identified at the first baseline as "high-event" drivers. CONCLUSIONS: Our results show that immediate visual feedback for adolescents and cumulative video feedback for parents and adolescents during the early period of independent driving can have a dramatic influence on the rate of safety-relevant driving events. To the extent that such events are a proxy for crash risk, we suggest that feedback can enhance adolescent driving safety.

Extending parental mentoring using an event-triggered video intervention in rural teen drivers.

Teen drivers are at high risk for car crashes, especially during their first years of licensure. Providing novice teen drivers and their parents with a means of identifying their risky driving maneuvers may help them learn from their mistakes, thereby reducing their crash propensity. During the initial phase of learning, adult or parental supervision often provides such guidance. However, once teens obtain their license, adult supervision is no longer mandated, and teens are left to themselves to continue the learning process. This study is the first of its type to enhance this continued learning process using an event-triggered video device. By pairing this new technology with parental feedback in the form of a weekly video review and graphical report card, we extend parents' ability to teach their teens even after they begin driving independently. Twenty-six 16- to 17-year-old drivers were recruited from a small U.S. Midwestern rural high school. We equipped their vehicles with an event-triggered video device, designed to capture 20-sec clips of the forward and cabin views whenever the vehicle exceeded lateral or forward threshold accelerations. Preliminary findings suggest that combining this emerging technology with parental weekly review of safety-relevant incidents resulted in a significant decrease in events for the more at-risk teen drivers. Implications for how such an intervention could be implemented within GDL are also discussed.