An Ensemble Learning Method for Robot Electronic Nose with Active Perception

The electronic nose is the olfactory organ of robot, which composed a large number sensors to perceive smell objects through free diffusion. Traditionally, it difficult realize active perception function, and meet requirements small size, low cost, quick response that robots require. In order address these issues, novel with was designed an ensemble learning method proposed distinguish different objects. An array three MQ303 semiconductor gas electrochemical sensor DART-2-Fe5 were used construct nose, four algorithms realized odor function. experiment results verified accuracy can reach more than 90%, even though 30% training data. based on improve efficiency data collection perception.