We present a model for visual behavior that can simulate the glance pattern observed around driver-initiated, non-critical disengagements of Tesla’s Autopilot (AP) in naturalistic highway driving. Drivers may become inattentive when using partially-automated driving systems. The safety effects associated with inattention are unknown until we have quantitative reference on how changes automation. is based data from 290 human initiated AP disengagement epochs. Glance duration and transition...

Electric scooters (e-scooters) are a relatively new and popular means of personal transportation in many cities. Unfortunately, they have been involved crashes with other road users whom share the infrastructure. Crashes motorized vehicles particularly critical since result more severe injuries or even fatalities. While previous work has highlighted consequences failed interactions, we know little about how drivers interact e-scooters to improve such interactions. In this paper, conducted...

Electric scooters (e-scooters) are a relatively new and popular means of personal transportation in many cities. Unfortunately, they have been involved crashes with other road users whom share the infrastructure. Crashes motorized vehicles particularly critical since result more severe injuries or even fatalities. While previous work has highlighted consequences failed interactions, we know little about how drivers interact e-scooters to improve such interactions. In this paper, conducted...

Many low-severity crashes with no injuries are not reported in crash databases due to specific sampling criteria. If this bias is addressed, safety assessment analyses could yield inaccurate results. Unreported cases an example of missing at random (MNAR) data. Recognizing and addressing MNAR data can be difficult; it requires domain expertise clear assumptions build accurate statistical models recover the complete sample. Previous studies have often created ad-hoc procedures arbitrary or...

This paper introduces a reference model of glance behavior for driving safety assessment. can improve the design automated and assistive systems. Technological limitations have previously hindered use unobtrusive eye trackers to measure in naturalistic conditions. presents comprehensive analysis eye-tracking data collected field operation test, using an tracker that proved be robust real-world scenarios. We describe post-processing technique enhance quality eye-tracker data, propose...

As new automated features enter the automotive market, we need methods to assess their safety in a rapid, proactive, and iterative way. The traditional way of relying on crash statistics does not meet these needs. An alternative is use extrapolation techniques designed deal with rare events, such as extreme value theory (EVT). In this paper, applied EVT estimate risk collision without adaptive cruise control (ACC) during steady-state car following. We defined Bayesian regression model...

Objective: Many low-severity crashes are not reported due to sampling criteria, introducing missing at random (MNAR) bias. If addressed, MNAR bias can lead inaccurate safety analyses. This paper illustrates a statistical method address such Methods: We defined custom probability distribution for the observed data as product of an exponential population and logistic reporting function. used modern Bayesian probabilistic programming techniques. Results: Using simulated data, we verified...

We quantify the time-course of glance behavior and steering wheel control level in driver-initiated, non-critical disengagements Tesla Autopilot (AP) naturalistic driving. Although widely used, there are limited objective data on impact AP driver behavior. offer insights from 19 vehicle owners when using transitioning to manual Glance were coded for 298 highway driving disengagements. The average proportion off-road glances decreased 36% was engaged 24% while manually after disengagement....

Although an automated vehicle may operate for extended time, it suddenly request a user's intervention in critical situations (i.e., beyond the system's operational design domain). Despite proliferation of studies to understand how users resume control such situations, systematic analysis whole response process is necessary. We analyzed visual-motor distracted front and lateral conflicts. also investigated effect false warnings expectations. In driving simulator experiment (high fidelity,...

Previous research indicates that drivers may forgo their supervisory role with partial-automation. We investigated if this behavior change is the result of time automation was active. Naturalistic data collected from 16 Tesla owners driving under free-flow highway conditions. coded glance location and steering-wheel control level around Autopilot (AP) engagements, driver-initiated AP disengagements, steady-state use in-between engagement disengagement. Results indicated immediately after...

Visual time-sharing (VTS) behavior characterizes an inattentive driver. Because inattention has been identified as the major contributing factor in traffic crashes, understanding relationship between VTS and crash risk could help reduce through development of countermeasures. The aims this paper are: 1) to develop a reference model 2) reveal if vehicle automation influences behavior. was based on naturalistic eye-tracking data. sequences were extracted from routine driving data (including...

Automated driving vehicles of the future will most likely include multiple modes and levels operation thus various transitions control (ToC) between human machine. Traditional activation devices (e.g., knobs, switches, buttons, touchscreens) may be confused by operators among other system setting manipulators also susceptible to inappropriate usage. Non-intrusive eye-tracking measures assess driver states (i.e., distraction, drowsiness, cognitive overload) automatically trigger...

Vision Zero postulates that no one should be killed or seriously injured in road traffic; therefore, it is necessary to define evidence-based speed limits mitigate impact severity.The overall aims guide the definition of safe speeds by establishing relations between and risk at-least-moderate (MAIS2+) atleast-severe (MAIS3+) injuries for car occupants frontal side crashes Sweden.As Swedish in-depth data are unavailable, first objective was assess applicability German In-depth Accident Study...

The landing phase is a critical stage in autonomous aerial landing, especially when the vehicle lands moving platform, as ground vehicles. In this paper, solution combining information from onboard camera of drone with an observer used to estimate and predict future position platform. This estimation control algorithm, based on quaternions, for generating tracking trajectory. proposed then validated real-time experiments (two scenarios) demonstrate well performance efficiency closed-loop...

Vision Zero postulates that no one should be killed or seriously injured in road traffic; therefore, it is necessary to define evidence-based speed limits mitigate impact severity. The overall aims guide the definition of safe speeds by establishing relations between and risk at-least-moderate (MAIS2+) at-least-severe (MAIS3+) injuries for car occupants frontal side crashes Sweden. As Swedish in-depth data are unavailable, first objective was assess applicability German In-depth Accident...