Two challenges for rescue robots are to detect human beings and to have an accurate positioning system. In indoor positioning, GPS receivers cannot be used due to the reflections or attenuation caused by obstacles. To detect human beings, sensors such as thermal camera, ultrasonic and microphone can be embedded on the rescue robot. The drawback of these sensors is the detection range. These sensors have to be in close proximity to the victim in order to detect it. UWB technology is then very helpful to ensure precise localization of the rescue robot inside the disaster site and detect human beings. We propose a new method to both detect human beings and locate the rescue robot at the same time. To achieve these goals we optimize the design of UWB pulses based on B-splines. The spectral effectiveness is optimized so the symbols are easier to detect and the mitigation with noise is reduced. Our positioning system performs to locate the rescue robot with an accuracy about 2 centimeters. During some tests we discover that UWB signal characteristics abruptly change after passing through a human body. Our system uses this particular signature to detect human body.