The work aims toexplore the design and analysis methods of human-machine interaction (HMI) in intelligent cockpits by extracting and analyzing human errors during the the driving process in response to the evolving trend of automotive intelligent cockpit systems shifting from a focus solely on driving to encompassing both driving and in-cabin experiences. Ten common non-driving tasks in intelligent cockpits were selected and error factors in their interaction processes were extracted through real-world operational experiments. By considering the cognitive mechanisms underlying these errors, clusters of factors contributing to errors in intelligent cockpits were established and the recurrence frequencies of error factors among three different cockpit interfaces were compared. The results indicated significant variations in error factor recurrence frequencies across different modules of the cognitive framework. Based on the cognitive theory, an in-depth deductive analysis was conducted on the interface design features contributing to these differences in error factors. In general, the incorporation of the human error analysis research methodology into the design research process of intelligent cockpit interfaces can significantly enhance the analysis of drivers' cognitive perceptions of interface information. It enables the evaluation of its impact on the driving experience, thereby offering valuable insights for human-machine interaction research in the field of intelligent cockpit design.