Estimating Object Location using Particle Clustering on Rotating Sonar Detection

Abstract

Particle filter been used for localization and position estimation with various applications. Several method applied in order to reduce the complexity of particle information to lower processing resource requirement while providing a precise map. Sonar sensor provide a range precision with poor bearing. Partial observation applied to gain estimation of object location using several measurement from multiple vantage point. This paper propose grouping of detection particle from sonar sensor using clustering method in order to provide estimated object position from a single vantage point. The approach estimate objects using euclidean distance treshold to separate one object detections from the other and group all the detection particle into cluster containing a set of number. As a result of particle clustering, object location can be estimated with their respective weight of certainty as an attribution for further decision making.