About me
Age-Driven Takeover Optimization in Automated Driving: Smart Wearables and Vibrotactile Feedback for Enhanced Safety - 1. Background
Automated systems are currently used in healthcare, manufacturing, and transportation. Older adults with cognitive and perceptual declines are expected to benefit from these systems. For example, automated vehicles capable of managing most driving tasks can assist with daily travel. However, the current automated system has limitations. When systems reach their limits, human intervention (i.e., takeover) is still required, which can be challenging for older adults. Therefore, developing an effective takeover request is crucial for ensuring a safe automated vehicle experience for older adults. Tactile takeover requests show promise, as the tactile channel may remain available while performing non-driving-related tasks. Previous research has examined parameters for efficient tactile warning displays, with wrist-worn devices preferred for their sensitivity and adaptability across vehicle models. Studies revealed that wrist devices offering dynamic vibrotactile feedback—vibrating sequentially at different locations—are more effective than static patterns for navigation purposes. However, driving simulator studies present conflicting results, suggesting that static patterns perform better than dynamic ones during the takeover. To address this conflict, this study aims to explore how directional vibrotactile feedback from wrist-worn devices impacts takeover performance across different age groups.
2. Method
The study employed a 2 (age: younger and older adults) × 4 (patterns: Baseline, Full-Dynamic, Semi-Dynamic, and Static) full factorial design. This study included forty participants, consisting of 20 younger adults and 20 older adults. During the experiment, participants were seated in a Level 3 automated vehicle traveling on a three-lane highway. When obstacles ahead led to system failures, a vibrotactile takeover request was sent through the wristband. After processing the request and evaluating the traffic conditions, participants are required to move into the lane that has the most space. There were a total of four drives in this experiment; each drive included one of four vibrotactile patterns. Both objective performance and subjective preference were adopted as dependent variables in this study. Reaction time (the time from request to deactivating automation) and takeover time (the time from request to first conscious action) were used to measure takeover performance. Usefulness and satisfaction were used to determine user preference regarding the vibrotactile patterns.
3. Result
The findings from the differences between vibrotactile patterns indicate that Baseline and Static patterns resulted in faster reaction and takeover times, as well as higher usefulness and satisfaction scores compared to the dynamic patterns. One explanation is that dynamic patterns take longer to convey complete information, requiring more time for driver interpretation. In contrast, static patterns offer immediate, recognizable vibrations that are more intuitive during takeovers. The analysis of age group differences reveals that older adults responded more slowly than younger adults, possibly due to age-related declines in perception and cognitive function. However, the takeover times were comparable for both age groups, indicating that increased driving experience might mitigate the variations in takeover duration. Overall, the study contributes to the inclusive design of human-machine interfaces for future automated vehicles.
