PROBABILISTIC FILTERING OF SONAR DATA

In this paper we propose an iterative probabilistic sonar filtering technique able to decide whether a sonar reading is good or not. This technique is applied to standard time-of-flight sonars, and may be used in real time applications. A non-iterative approach based on a voting schema is also discussed, reducing even more the computational cost. The technique does not depend on any specific mapping, localization or navigation technique. Thus, a large range of applications is possible. A quality metric to evaluate the goodness of sets of raw readings is also presented. Experimental results demonstrate the fast convergence of the algorithm, as well as its robustness and the quality of the filtered readings.