After a significant earthquake, most deaths are caused by buildings collapsing and trapping victims under rubble. Rapid localization of injured survivors by rescue teams to prevent death is a major issue. Considering the time limitation and the safety of the rescue team, a new search and rescue (SAR) method which could rapid localize the survivors, avoid direct access and exploration of the affected area is required. The objective of this study is providing a new search and rescue method by using an autonomous sensor system to detect the victims under rubble.  We developed a new life-detection sensor system includes a gas sensor, air condition monitoring sensor, microphone, and an RF communication module.  The gas sensor shows the casualty position. The air condition monitoring sensor shows the credibility of the gas sensor. The microphone records the voice and we developed the algorithm for recognizing the voice source. To solve the communication problem under rubble, the RF mesh network had been implemented in our system. On the software side, the GUI can show the raw sensor data, estimate the victim position based on gas sensor data and show the suspect position by the color map.