In recent years, the demand for robots has increased, and their roles have changed. In other words, there are increasing opportunities for them to be used not only in industrial applications such as factories, but also in daily life. In order for users to continue using these robots, the impression they have towards the robots needs to be considered. However, at present in the new coronavirus disease, it is difficult to actually assemble a robot. Instead, designing and developing a robot in virtual space can be an alternative. In this study, we evaluated the affective values of robots in a virtual space. We created a virtual space as a university campus, along with three pairs of robots with different shapes. Then, we performed kansei evaluation of the robots employing Semantic Differential (SD) for questionnaire. The results show how the ratings differ for each of the robot pairs and adjective pairs. In particular, we found that some adjective pairs had higher ratings than other adjective pairs, suggesting different impressions on our designed robots.