자율주행 이동 공간에서의 탑승 경험과 정서 반응 연구 - 멀티모달 행동 관찰을 중심으로 -

초록

(Background and Purpose) Driven by the rapid advancement of autonomous driving technology, vehicle interiors are fundamentally shifting from driving-centered functional environments to interactive mobility spaces where continuous interaction between passengers and autonomous systems occurs. As vehicle control shifts completely from human drivers to automated systems, distinct affective responses, such as limited situational awareness, profound loss of control, and technology distrust, may emerge among passengers. This phenomenon highlights the academic and practical imperative to reinterpret the vehicle interior as a strategic design environment in which passenger experiences and affective responses are constructed. Accordingly, this study aimed to analyze passengers' multimodal behavioral responses within automated interiors, examine their structural relationships with affective dimensions, and derive implications for user-centered interaction design. (Method) This study systematically adopted a multimodal behavioral analysis framework integrating empirical experimental research with artificial intelligence(AI)-assisted behavioral observations. Within a simulated autonomous-driving interior environment, unpredictable scenario-based tasks—including sudden stops, route modifications, stop requests, and system non-responses—were established, and the think-aloud protocol was used to capture participants' real-time behaviors and verbal expressions. A generative pre-trained transformer (GPT)-based multimodal AI model was used as an auxiliary analytical tool for the initial classification of the behavioral data. The final interpretations were then derived through researcher-led behavioral coding and systematic triangulation of researcher observations, questionnaire data, and in-depth interviews. (Results) The empirical findings demonstrated that environmental exploration behaviors and verbal responses served as core behavioral indicators of limited situational awareness. Meanwhile, self-comforting gestures and explicit system control requests were significantly associated with a subjective loss of control. Conversely, participants produced limited verbal responses aimed at verifying or validating the automated system's judgments and responses. This specific characteristic was interpreted as a constructive behavioral attempt to comprehend the system's decisions rather than representing definitive evidence of technology distrust. Furthermore, the self-report measures and interview outcomes strongly supported these behavior-based interpretations, confirming the interpretive validity of the multimodal approach. (Conclusions) This study reinterprets the autonomous vehicle interior as a complex mobility space shaped by passengers' spatial experiences and applies a multilayered analytical approach integrating multimodal behavioral responses, affective responses, and self-report data to interpret passenger experience. Based on the behavioral characteristics associated with limited situational awareness, loss of control, and technology distrust, the findings provide explicit guidelines for user-centered interaction design, including situational information support, explanatory structures for system responses, and adaptive feedback.

키워드