This paper presents the results of tests of interactions between real humans and simulated vehicles in a virtual scenario. Human activity is inserted into the virtual world via a virtual reality interface for pedestrians. The autonomous vehicle is equipped with a virtual Human-Machine interface (HMI) and drives through the digital twin of a real crosswalk. The HMI was combined with gentle and aggressive braking maneuvers when the pedestrian intended to cross. The results of the interactions were obtained through questionnaires and measurable variables such as the distance to the vehicle when the pedestrian initiated the crossing action. The questionnaires show that pedestrians feel safer whenever HMI is activated and that varying the braking maneuver does not influence their perception of danger as much, while the measurable variables show that both HMI activation and the gentle braking maneuver cause the pedestrian to cross earlier.

翻译：本文展示了真实人类与仿真车辆在虚拟场景中交互的测试结果。通过为行人构建的虚拟现实界面，人类活动被无缝嵌入虚拟世界。自动驾驶车辆搭载虚拟人机界面（HMI），在真实人行横道的数字孪生场景中行驶。当行人意图过马路时，HMI结合了平缓与激进两种制动策略。交互结果通过问卷调查及可量化变量（如行人开始穿越动作时与车辆的距离）获取。问卷调查表明，HMI启动时行人安全感显著提升，而制动策略的变化对其危险感知影响较小；同时可量化变量显示，HMI启动及平缓制动策略均促使行人更早完成穿越行为。