A8.3 - Improvements on the Robustness of Ultrasonic Bearing Measurement Methods

SENSOR+TEST Conferences 2011
2011-06-07 - 2011-06-09
Proceedings SENSOR 2011
A8 - Ultrasonic Sensors III
C. Walter, G. Kaniak, H. Schweinzer - Vienna University of Technology (Austria)
193 - 198


Scene analysis has become an important part of ultrasonic research due to its broad applicability in industrial applications, robotics, supervising and other fields. Some reasons why ultrasonic analysis is so attractive for these applications is the inherent reduction of data, high insensitivity to dust or foggy atmosphere and low cost of the sensors. Over time a lot of different methods have evolved which are distinguished by the number of sensors, by targeting 2D or 3D scene analysis and by the type of additional information they employ like distance, object type and spherical coordinates.

One of the methods we have developed is special in the way that it requires only a single transducer but still is able to determine the spherical coordinates of an object. It applies a technique that uses broadband signals to control the radiation pattern of the transducer. The received echoes are then analyzed using an especial algorithm and the distance, the polar angle and the azimuth angle are returned as a result. Measurements have shown that accuracy can be better than ±1°for the polar angle (for 98% of all values) and ±0.1mm (2σ with temperature compensation) for the distance.

This article shows extensions of our previous work. We have upgraded our primary algorithm to work with multiple reflections which can be partially overlapping. As a first step we have removed the necessity of choosing and cutting a time window from the received input signal. For that purpose the algorithm is modified such that it does not only use the peaks in the correlated input signal but also their relative displacement. For objects in different directions and with approximately the same distance from the transducer, echo overlapping becomes a crucial factor. Thus, we have investigated the effects of echo overlapping on the performance of our algorithms and are outlining our future research work on echo separation.

As a second improvement we have worked on increasing the robustness to noise in the signal. This has been a problem especially higher polar angles were the energy of acoustic waves becomes significantly reduced.