Preferences of Individuals With Different Levels of Visual Impairment for Autonomous Vehicles

Abstract

Autonomous vehicles hold promise for enabling independent travel for individuals with visual impairment. This study examined the relationship between severity of visual impairment and preferences for autonomous vehicles. Individuals with visual impairment completed an online survey with questions about their visual status, their trust and acceptance for vehicles of different automation levels, and their preferred modes of interaction with the vehicles. The results showed a general preference for vehicles with full automation. However, individuals with less severe impairment showed considerable trust and acceptance for lower levels of automation. While most individuals prefer the auditory/speech as the modality of interaction with autonomous vehicles, those with less severe impairment still wish to use the visual modality of interaction. Our findings suggest that preferences for autonomous vehicles vary depending on the severity of visual impairment, which should be considered during the design of the vehicles.

Autonomous vehicles hold great potential for enabling independent travel for individuals with visual impairment. Previous studies suggested that these individuals were generally hopeful about the technology but also expressed concerns about the accessibility of vehicle features (Bennett et al., 2020; Brinkley et al., 2020). Visual impairment comes in different forms and can range from partial vision loss to total blindness. It is currently unclear how the severity of an individual’s visual impairment might affect their trust and accessibility-related preferences for autonomous vehicles. This study examined the relationship between the severity of visual impairment and the preferences for two aspects of autonomous vehicles: level of automation and modality of interaction.

Individuals with visual impairment completed an online survey and answered questions about their visual status, their trust and acceptance for vehicles of different automation levels, and their preferred modalities of interaction with the vehicles. Three levels of automation were included: low, high, and full automation, roughly corresponding to SAE levels of 3 to 5. Before answering questions about vehicles of an automation level, the participants read a scenario depicting the features and limitations of that level. A few questions adapted from an existing questionnaire on trust in automation (Jian et al., 2000) were used to measure respondents’ trust for the vehicle depicted in the scenario on a 7-point scale. One question was asked about their perceived ability to operate the depicted vehicle (“I should be able to operate this car”) and another question about their acceptance/intention to own the vehicle (“I would like to own a car like this”). The respondents were then asked to select their preferred display modalities (visual, auditory, or tactile) for different types of trip-related information (e.g., time to destination, current location, detour). The respondents were also asked to rate their comfortability level (on a 5-point scale) with using each of the following modalities to control/send commands to the vehicle: eye movement; BMI (brain-machine interface); mid-air gesture; speech; visual/touch screen.

Based on self-reported visual status, 27 valid responses were divided into two groups: Blind (n = 13) and Low Vison (n = 14). The respondents generally preferred the highest level of automation (level 5) and their trust increased as the automation level increased. Whereas the two groups had similar trust and acceptance for level 5 vehicles, the Low Vision group had higher trust and acceptance for level 3 and level 4 vehicles than the Blind group. The respondents’ trust for vehicles of a certain automation level correlated with their perceived ability to operate the vehicle (e.g., r = .72 for Level 4). The respondents (90%) dominantly preferred an auditory display for the delivery of trip-related information. In addition to an auditory display, many respondents (30%) also preferred a visual display. The tactile display was the least preferred. The respondents showed the strongest preference for speech/voice as the modality of vehicle control. The Low Vision group showed higher preference for the visual/touch modality of control, compared to the Blind group. Preferences for other control modalities were generally low.

Our findings suggest that preferences for autonomous vehicles vary depending on the severity of visual impairment, which should be considered during the design of autonomous vehicles. While individuals with visual impairment generally prefer level 5 vehicles, those with less severe impairment showed considerable trust and acceptance for level 4 vehicles, presumably due to their relatively high perceived ability to operate vehicles of this automation level as well as a desire to retain certain level of control over the vehicle. While most individuals prefer the auditory/speech as the modality of interaction with autonomous vehicles, a considerable percentage of individuals, particularly those with less severe impairment, still wish to use the visual modality of interaction. The design of future autonomous vehicles should aim to facilitate the use of their residual vision for users with visual impairment.