Processing of Ultrasonic Echo Envelopes for Object Location With Nearby Receivers

Ultrasonic sonar systems are commonly used for obstacle location in robotics and autonomous navigation applications. When irregular objects have to be located, optimal algorithms for time-of-flight (TOF) estimation, like cross correlation, might be unreliable since complex reflecting surfaces can destroy phase coherence during the duration of the echo. Direct threshold methods acting on the envelope are equally inaccurate. This paper proposes a simple digital processing technique, called selective normalization, that works with the signal envelope and robustly determines the differential TOF (DTOF) of the ultrasonic echoes received by two or more nearby transducers. It is empirically demonstrated that this method improves the positioning accuracy of nonpointlike objects.