Development of visualization system for dynamic gas/odor behavior-Realization of video camera for gas distribution

由于恶臭、污染、爆炸物的来源需要迅速被发现，因此开发用于定位气味来源的传感系统非常重要。然而，目前还没有开发出具有足够性能的人工系统。由于气味一般由风携带，并伴有湍流，因此其流动行为非常复杂。虽然我们小组之前开发的羽流追踪机器人可以在洁净室等相对恒定流量的房间中搜索源，但是对于目前的机器人来说，在室外场地或几乎没有风的房间中进行可靠的探测是一项困难的任务。然后，引入均匀传感器阵列来可视化气体/气味的动态行为，而不是以前使用的几种传感器。当我们使用该传感器阵列时，可以利用增加方向估计可靠性所必需的冗余信息来抑制湍流的影响。首先，通过仿真优化了传感器阵列结构，获得了清晰的图像。其次，对脉冲驱动半导体气体传感器阵列进行了实验。但是，由于响应的恢复时间不够，只能在开始喷射蒸汽或停止蒸汽暴露时获得流动方向。最后，将半导体气体传感器阵列替换为时间常数约为1秒的小型化石英晶体微天平气体传感器阵列。利用该传感器阵列，不仅在风洞内，而且在风速变化比风洞大得多的洁净室，都能很好地实现气流的可视化和方向估计。

It is important to develop the sensing system for localizing an odor source since the sources of bad smell, contamination, and explosive should be rapidly found. However, the artificial system with sufficient capability has not been so far developed. Since the odor is generally carried by wind accompanied by turbulence, the behavior of its flow is very complicated. Although the robot for plume tracing previously developed by our group is available for searching the source in a room with comparably constant flow such as clean room, the reliable exploration in the outdoor field or in the room almost without the wind is difficult task for the present robot.Then, a homogeneous sensor array was introduced to visualize the dynamic gas/odor behavior instead of a few sensors previously used. When we use that sensor array, the redundant information indispensable for increasing the reliability of the direction estimation is available to suppress the influence of the turbulence.First, the sensor array configuration was optimized to obtain a clear image by the simulation. Next, the experiment on the pulse-drive semiconductor gas sensor array was performed. However, the flow direction was obtained only at the starting time for spouting vapor or at the time for stopping the vapor exposure since the recovery time from the response was not sufficient.Finally, the semiconductor gas sensor array was replaced by that of miniaturized QCM (quartz crystal microbalance) gas sensors with time constant around a second. Using that sensor array, the gas flow can be always visualized clearly and the estimation of the flow direction was successful not only inside wind tunnel but also in the clean room where the variation of the wind speed was much larger than that in the wind tunnel.