One of the greatest problems in rescue operations during fire disasters is the blocking of firefighters’ view by dense smoke. Assuming that a firefighter’s most important task is to understand the situation within a smoke-filled space. We developed a way to do so, starting by scanning space using millimeter-wave radar combined with a gyrosensor. To detect persons and objects, we constructed a 3D map from signal reflection datasets using 3D image processing. We detail our proposal and report results of measurement experiment in actual smoke-filled areas.