Visual attention and steering wheel control: From engagement to disengagement of Tesla Autopilot

Abstract

Previous research indicates that drivers may forgo their supervisory role with partial-automation. We investigated if this behavior change is the result of the time automation was active. Naturalistic data was collected from 16 Tesla owners driving under free-flow highway conditions. We coded glance location and steering-wheel control level around Tesla Autopilot (AP) engagements, driver-initiated AP disengagements, and AP steady-state use in-between engagement and disengagement. Results indicated that immediately after AP engagement, glances downwards and to the center-stack increased above 18% and there was a 32% increase in the proportion of hands-free driving. The decrease in driver engagement in driving was not gradual over-time but occurred immediately after engaging AP. These behaviors were maintained throughout the drive with AP until drivers approached AP disengagement. In conclusion, drivers may not be using AP as recommended (intentionally or not), reinforcing the call for improved ways to ensure drivers’ supervisory role when using partial-automation.

(Mis-)use of standard Autopilot and Full Self-Driving (FSD) Beta: Results from interviews with users of Tesla's FSD Beta

“…Respondents indicated becoming complacent over time, taking their eyes off the road for relatively long stretches (e.g., 44 miles without looking up), and engaging in secondary activities (e.g., working on a laptop). This concurs with other studies in partially automated driving reporting drivers taking eyes off the road (Solís-Marcos et al, 2018;Morando et al, 2021) and engaging in secondary activities (Endsley, 2017;Banks et al, 2018;Wilson et al, 2020;Kim et al, 2021;Metz et al, 2021). An on-road study by Stapel et al (2022) did not reveal substantial differences in driver attention between manual driving and partially automated driving.…”

“…Our study found evidence of one-handed, hands-free, and mindoff driving. Studies have provided empirical evidence that users of Autopilot provide low or even no steering wheel control (Banks et al, 2018;Kim et al, 2021;Morando et al, 2021). Our study revealed that the sensitivity of Autopilot's torque steering wheel sensor might encourage one-handed driving, which represents a violation of recommended use according to Tesla's driver manual, requiring drivers to keep their hands on the steering wheel at all times (Tesla, 2022a).…”

Visual Attention Patterns in Automated and Manual Lane Change Maneuvers

“…In conclusions, this work sought to connect previous work on driver glance behavior related to automation use -specifically Tesla Autopilot (Gaspar & Carney, 2019;Morando et al, 2021)-and the work highlighting the importance of visual attention in lane change maneuvers (Chovan et al, 1994;Fitch & Hankey, 2012;Shinar, 1978;Wang & Knipling, 1994). Differences in driver glance behavior observed between lane changes performed manually and those which used automation show patterns that suggest that drivers may allocate their visual attention differently depending on the automation state of the vehicle and the ability of the driver to anticipate the maneuver.…”

Modeling Driver-Automation Interaction using A Naturalistic Multimodal Driving Dataset

“…This paper focuses on an exploratory analysis of a subset of the study data encompassing a randomly sampled set of 50 trips from Driver A and 48 trips from Driver B. The two male drivers, 35 and 40 years old at study enrollment, were chosen according to an initial analysis of glance behavior derived from Morando et al, 2021. For Driver A, 44% of Autopilot epochs (one epoch here means a 30-s driving period) were dominated by off-road glances (which occupied more than 50% of an epoch). In contrast, Driver B concentrated on the road in 96% of Autopilot epochs.…”

“…At each moment, vehicle modemanual or automatedcan be maintained or changed according to driver motivation and state, vehicle dynamics, and driving environments. However, few studies have explored aspects of automation use during steady state (e.g., Morando et al, 2021), or the factors impacting the moment-tomoment driver-automation collaboration in naturalistic environments. This type of research will expand our understanding of drivers' holistic use of automation and provide a reference that can aid in developing more comprehensive behavioral and safety assessments.…”