Mobile devices have become wide spread, as a result of improvements in hardware and communication technologies. These devices are equipped with some sensors to protect themselves. In recently years, technologies have been developed that allow a device to recognize its environments using sensor data are developed. However, in an indoor environment, there are objects that affect sensor data. Therefore, sensors required either calibration to adjust their behavior, or clarification of the cause of incorrect behavior. This requires large amounts of sensor data to be collected indoors, but it is difficult for humans to collect this data in the indoor environment. Methods of data collection by humans are limited by the number of observation points that can reasonably be attained. In this paper, we proposed the Fine Mesh method for dense sampling using an autonomous robot. Our proposed method aims to increase the number of observation points. We constructed an autonomous robot that collects signals at 10 cm intervals using positioning estimates with camera images. To confirm the effect of collecting signals in more detail, we evaluate the accuracy of a Wi-Fi indoor positioning system constructed by our robot. We then use our platform to evaluate the efficacy of detailed signal collection.