Network Model Based Odor-Source Searching Robot

Neural network model was reconstructed based on morphological and physiological properties of brain neurons related to pheromonal response called flip-flip, which is thought to be a command signal for odor-source localization of silkmoths. This network model was implemented to a small mobile robot. The size of the robot is similar to that of the silkmoth 31 mm (L) x 18 mm (W) x 30 mm (H). Because odor sensors sensitive to pheromone were not developed, insect antennae isolated from a silkmoth were used as pheromone sensors of the robot. Olfactory responses were recorded by placing electrodes to the base and tip of the antenna, which is called an electro-antenno gram (EAG).  Then the behavior of the robot was evaluated in a wind tunnel. EAGs were transmitted to an external computer where the network model was implemented. Processed data was returned to control the robot movement. In response to a pulsed pheromonal stimulation, the robot showed a typical movement pattern consisting of a surge, zigzag turns and a looping. When the robot was placed at downwind of the pheromone in the wind tunnel, the robot repeated the typical patterns and succeeded in orienting toward the pheromone source. This robot was the first robot which could find out the pheromonal odor.

updating of the site
Copyright (C) 2018 Neuroinformatics Unit, RIKEN Center for Brain Science