Over the last years, driving automation has increasingly moved into focus in human factors research. A large body of research focusses on situations in which the human driver needs to regain control. However, little research has so far been conducted on how SAE level 3+ automated driving should be designed with focus on occupant comfort. This thesis aims at identifying a comfortable driving style for automated vehicles. As a basis, it was necessary to pinpoint driving metrics, which vary between driving styles and can be manipulated in order to design a comfortable driving style. Hence, Study 1 was conducted, in which drivers (N = 24) manually drove on a highway or on urban and rural roads with certain driving styles. Results show relevant metrics (i.e., lateral and longitudinal acceleration, lateral and longitudinal jerk, quickness, and headway distance in seconds) and that these metrics vary across maneuvers and thus, a maneuver-specific analysis is recommended. As these metrics are derived from manual data, it remained unclear after Study 1, in which range the metric values should vary for comfortable automated driving. Therefore, as a second step, the main metrics were varied and the subsequent combinations implemented in an automated vehicle as well as in a dynamic simulator with two different configurations. The combinations were then subject to ratings by 72 participants. Results show that the metrics and values found in Study 1, are able to elicit a range of comfort ratings in automated driving. It was also found, that acceleration is a key variable in experiencing comfort. However, it is not the sole predictor. Additionally, as higher levels of automated driving with larger velocities are still bound to considerable constraints for on-road testing, the second study was also used to validate a dynamic driving simulator to allow comfort during automated driving to be studied. In comparison to ratings on a test track, the dynamic simulator setting with longitudinal orientation is able to show both relative and absolute validity of comfort ratings. In the third and final step, different approaches to automated maneuvers were rated by participants (N = 72) regarding the comfort they experienced. A lane change, an acceleration, and a deceleration maneuver were chosen as test maneuvers. The lateral or longitudinal acceleration was varied in each of these maneuvers. Results, again, show comfort ratings are maneuver specific. On one hand, symmetrical and early-onset lane change maneuvers and symmetrical acceleration maneuvers were preferred. However, symmetrical deceleration maneuvers and deceleration maneuvers with a slower acceleration decrease evoke the highest comfort ratings. These ratings made it possible to offer guidelines for the design of automated driving styles. Furthermore, dependence on a number of personality traits was analyzed. Results suggest the general preference for certain driving styles to be unaffected by personality. However, it seems, participants with certain personality types are less particular about their preference for certain driving styles. Summed up, comfortable automated driving is – under the investigated circumstances – characterized by maneuvers with sufficient headway distance and smooth applications of small acceleration and small jerk. These should, even so, still provide sufficient motion feedback. Surrounding traffic seems to play an important role through urgency and should be considered for on-road implementation. Differences in personality did not seem to play a crucial role.
Comfort in Automated Driving: Analysis of Driving Style Preference in Automated Driving
2017-09-21
Hochschulschrift
Elektronische Ressource
Englisch
Comfort in automated driving : analysis of driving style preference in automated driving
TIBKAT | 2018
|Europäisches Patentamt | 2022
|AUTOMATED DRIVING SYSTEM, AUTOMATED DRIVING APPARATUS, AND AUTOMATED DRIVING METHOD
Europäisches Patentamt | 2020
|AUTOMATED DRIVING DEVICE AND AUTOMATED DRIVING METHOD
Europäisches Patentamt | 2022
|Europäisches Patentamt | 2019
|