在测量驾驶风格时，克服各种量化参数和驾驶者行为的不可预测性是一项挑战。各种特征选择、分类算法和驾驶员分类的变化已经被有关研究认为是解决这些问题的潜在答案。然而，对于自动驾驶来说，驾驶员的情况会发生了变化，同样的特征和标准不能直接拿来预测自动驾驶的驾驶风格。通常情况下，风格预测的目标是使理想的驾驶风格与特定驾驶员相匹配，从而使自动驾驶系统更加自然。实现这种匹配将加快自动驾驶系统的实施，提高安全性，并增加自动驾驶的便利性和自由度。这篇硕士论文研究了在高级驾驶辅助系统（ADAS）和自动驾驶（AD）系统 导入驾驶风格时各种驾驶特征的重要性。该研究评估了使用特定特征进行驾驶风格预测的准确性，探讨了自动驾驶中驾驶员对不同驾驶风格的偏好，以及人格特征对驾驶特征和偏好驾驶风格的影响。主要研究结果表明，某些驾驶特征，如偏航率和方向盘速度，在三种不同的驾驶情况下更能预测驾驶风格，而其他特征如车速则不太重要。基于所选驾驶特征的驾驶风格预测准确率是适中的，解释的变异率在 10%到 34%之间。论文还显示，驾驶员的个人多维驾驶风格清单（MDSI）因素对他们在自动驾驶中的喜好驾驶风格没有显著影响，这表明在预测自动驾驶风格时，个人手动驾驶风格的重要性较低，对自动驾驶的信任和对安全的感知起到了更大的实质性作用。此外，研究结果显示，人格特征，特别是一致性和透明性，影响自动驾驶中的驾驶风格喜好，并与各种场景下的驾驶行为相关。研究结果强调了自动驾驶系统中的驾驶特征、个性特征和驾驶风格偏好之间的复杂关系，凸显了在开发用户友好型自动驾驶技术的过程中，了解和迎合个体驾驶者的偏好和信任水平的重要性。未来的研究应进一步探索人格特征与驾驶行为和偏好之间的联系机制，并考虑增加驾驶特征或生理指标来提高人格预测准确性，而不是采用手动驾驶风格来预测。
When measuring driving style, overcoming the variety of quantitative parameters and the unpredictability of driver behavior is challenging. Various feature selections, classification algorithms, and driver categorization variations have been examined as potential answers to these problems. However, for autonomous driving, the driver‘s circumstances change, and the same traits and standards cannot be directly used to apply the driver‘s driving style. Typically, the goal is to match the driver with the ideal driving style to make autonomous driving systems more natural. Achieving this fit would expedite the implementation of autonomous driving systems, improve safety, and increase the convenience and freedom of autonomous driving.This master thesis investigates the significance of various driving features in implementing driving styles into Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) systems. The study assesses the accuracy of driving style prediction using specific features and explores the drivers‘ preferences for different driving styles in autonomous driving, along with the influence of personality traits on driving features and preferred driving styles. Key findings indicate that certain driving features, such as yaw rate and steering wheel speed, are more predictive of driving styles in three distinct driving situations, while other features like vehicle speed are less significant. Driving style prediction accuracy based on the chosen driving features was found to be moderate, with explained variances ranging from 10 to 34 percent.The thesis also reveals that drivers‘ personal Multidimensional Driving Style Inventory (MDSI) factors do not have a significant impact on their preferred driving style within autonomous driving, suggesting that the personal manual driving style has less importance and that trust in autonomous driving and perception of safety play a more substantial role. Furthermore, the results show that personality traits, particularly Agreeableness and Conscientiousness, influence the preferred driving style in autonomous driving and are correlated with driving behavior across various scenarios. The findings highlight the complex relationship between driving features, personality traits, and driving style preferences in autonomous driving systems, emphasizing the importance of understanding and catering to individual drivers‘ preferences and trust levels in the development of user-friendly autonomous driving technologies.Future research should further explore the mechanisms linking personality traits to driving behavior and preferences and consider additional driving features or physiological indicators to improve personality prediction in vehicles instead of manual driving style prediction.
