Skip to Main Content
This article describes a natural landmarks detection method to use with conventional 2D laser rangefinders. The method consists of three main parts: data clustering, smoothing and segmentation. A smoothing algorithm within a scale space framework is introduced to smooth the range image. This is achieved by repeatedly convolving the scan data with an adaptive smoothing mask calculated according to the Mahalanobis distances from a curve-based estimator, which tracks the features using UKF (unscented Kalman filter). Clustered data is segmented and characterized by the curvature of the range data. This method is robust to noise, and can reliably detect landmarks in the unstructured environment. Experimental results show that the proposed method is efficient in natural-landmark extraction.