Appropriate human likeness for social robots is said to increase trust and acceptance. Whether this applies to human communication features like dialog initiative needs to be investigated. Dialog initiative could be unacceptable for a robot, depending on the dialog content. Hence, the presented study investigates how a social robot’s proactive verbal and non-verbal communication behavior affects trust and acceptance depending on dialog content and content presentation order. A laboratory study (n = 31) with a humanoid robot was conducted. Talk initiative (human/robot) and the robot’s gaze behavior (directed/random) were manipulated. Dialog content was alternated between a service task and small talk. The subject’s trust, acceptance and human-robot proximity were assessed. Whereas a directed gaze was perceived as more humanlike and was more accepted during small talk, no gaze preference for the service task emerged. There was no preference for who initiated the small talk but for the service task, robot initiative led to higher trust in the robot when the service task was the first interaction. Participant’s self-reported trust in the robot was associated with the distance they kept to the robot. Different gaze and proactive strategies seem to be efficient to foster trust and acceptance in social robots for different dialog contents and thus should be considered when designing interaction strategies for social robots.

据说社交机器人适当的人像可以增加信任和接受度。这是否适用于诸如对话主动性之类的人类交流功能，需要进行调查。对话框的主动性对于机器人可能是不可接受的，具体取决于对话框的内容。因此，本研究调查了社交机器人的主动言语和非言语交流行为如何根据对话内容和内容呈现顺序影响信任和接受度。一项实验室研究（ñ= 31）用人形机器人进行。说话主动性（人/机器人）和机器人的注视行为（定向/随机）得到了控制。对话内容在服务任务和闲聊之间交替显示。评估对象的信任度，接受度和人机接近性。有方向的凝视被认为更人性化，并且在闲聊时更容易被接受，但没有凝视对服务任务的偏好。没有人愿意发起闲聊，但对于服务任务，当服务任务是第一次交互时，机器人主动性会导致对机器人的更高信任。参与者对机器人的自我报告信任与他们与机器人保持的距离有关。