The work aims to select the typical scene of a vehicle coming from behind in autonomous driving to explore the influence of transparency-based human-machine interface (HMI) design on the driver's cognition. Literature review, scene analysis, experimental evaluation and other research methods were adopted to design the HMI in this scene by the Situation Awareness-based Agent Transparency (SAT) model, and the design was verified on a driving simulator. The results indicated that the HMI transparency level was positively correlated with the driver's trust in the system, but the perceived workload had no significant change. In addition, different levels of transparency affected the drivers' situational awareness ability, and the drivers' understanding of the current situation was significantly correlated with the trust. The transparency design of HMI will affect the drivers' cognition from three aspects:situation awareness, workload and trust. Therefore, it is necessary to focus on the influence of these factors when designing the transparency of human-machine interface.