Robot with olfactory sense

点击翻译按钮获取中文摘要

A robot with olfactory sense to search for the source of gas or odor is expected to be applied to the gas leak detection, the early discovery of the fire, the detection of the danger in the mine etc. Although the previous research revealed that the robot with the gas and anemometric sensor could search for the gas/odor source, it took much time to explore the source due to the slow response/recovery speed of the semiconductor gas sensor. In the present study, the improvement of the algorithm for detecting the variation of the sensor response enabled the speedup seven times. Furthermore, the quartz resonator gas sensor with faster response/recovery speed was used. The improvement of the measurement-circuit time resolution and the baseline stability by compensating for the slight variation of the humidity enabled the gas detection even during the rapid movement of the robot (20cm/s). Further speedup may be achieved when the quartz resonator gas sensor is used.The algorithm for searching for the target can be studied in detail if the simulation method is established. Then, the simulation tool for estimating the sensor response when it is virtually placed in the plume. The transient sensor response model composed of a neural network was effective to estimate the sensor response where the vapor with the concentration rapidly changing within a second was exposed to it. Using that tool, the method for obtaining the distribution image of the gas concentration by minimum number of the gas sensors was studied. It was found that the sparse placement of the sensors along the wind direction in spite of the dense placement across the wind flow gave us the minimum information loss of the image. The developed tool may enable the simulation for searching for the gas/odor source.