Determining Voice Control Boundaries for in-Vehicle Functions: A User-Engineer-Expert Assessment Framework

Efficiently adapting voice control functions in smart vehicles remains a practical challenge in human-computer interaction. This study established a three-source, lightweight research system comprising user questionnaire survey (N = 326), engineer evaluation from the corporate sector (N = 12), and expert interviews from the industry (N = 8). A six-dimensional weighted adaptability assessment model(SWAAM) was designed (safety risk 30%, operational complexity 20%, usage frequency 15%, command clarity 15%, feedback requirements 10%, and technical feasibility 10%). Within 15 days, a systematic evaluation of 30 vehicle control functions was completed. The results indicate that user voice preference exhibits significant stratification: preference rates for high-frequency comfort functions such as navigation, air conditioning, and media exceed 98%, while preference rates for safety-critical functions such as driver assistance and chassis control are below 60%. Through triangulation validation, 15 items were identified for the whitelist (priority for release), 3 for the blacklist (prohibited from release), and 12 for the graylist (requiring scenario-based release, e.g., window control limited to speeds below 80 km/h, special scenario modes restricted to parking only). Expert consensus established safety red lines and design specifications, including “prohibition of voice control for core driving functions,” “adoption of the “action + object” command format,” and “requirement for secondary confirmation for safety-related functions.” The weighted total score across the six dimensions showed high consistency with the functional list (whitelist ≥ 7 points, blacklist <4.0 points), validating the model’s effectiveness. This method ensures scientific rigor while reducing the adaptation cycle to 15 days. It provides a practical, lightweight framework for rapid iteration in mass-production voice interaction projects.