G1500201 災害作業ロボットの障害物回避のためのセンシング技術

抄録

To prevent the outflow of dangerous materials in emergency, valve closing is an important task at large-scale plants such as natural gas. However, it is a challenging task to correspond the same time in the accident at plural points from a huge site and lack of specialists at large-scale plants. In addition, correspondence among workers becomes complex when there are risks to be connected for casualty that human error assumes a cause, the second disaster by occurring fire accidents. Therefore, valve closing using a robot has been required. The purpose of this study is to develop an autonomous mobile control application used for a disaster working robot. This paper presents an object detection method to judge valves that conduct it from a visual remarkable degree by autonomous work. We consider that the degree is similar than that of using template matching used conventionally improved by our comparing experiment.