2017 年 83 巻 853 号 p. 17-00108
Searching an odor source is a challenging task for a distributed autonomous mobile robot due to the complexity of odor formation in the air. Because the shapes of odor plumes highly depend on the environmental conditions including dynamical changes of wind direction, a robust performance of searching is highly demanded. In this paper, we introduced an olfactory active sampling device that was equipped electronic fan to generate air current and suction olfactory information to the sensors. Our main contribution was proposing and evaluating adaptive sampling strategies for appropriate odor-searching behavior in real environments by using the device. The active sampling device was designed after observations of the wing-flapping-effect of a male silkworm moth, which corresponding to‘ sniffing ’of mammals. The device was composed of an electronic fan, two (left and right) gas sensors embedded in air ducts, and was attached to a mobile robot. Each duct had a flange so as to take the air in front of a robot effectively, and the electric fan attached on the end of the duct excluded odors incoming from the rear. We settled different environmental conditions in which the fluctuation of the air flow was changed. We conducted experiments to verify the feasibility of the adaptive sampling system. As a result, under most of the environments, the robot with active sampling showed better performance than that without active sampling. However, the result showed that optimal active sampling has environmental dependency. To realize more adaptive system, we suggested that it is necessary to change the activeness of inspiration depending on the situation.
