Uncertainty of Line Segments Extracted from Static SICK PLS Laser Scans A. Diosi and L. Kleeman Uncertainty of Line Segments Extracted from Static SICK PLS Laser Scans

Data fusion using Kalman filters requires reasonably good error models. Our intention to fuse line segments, corners and edges obtained from a laser scanner and from advanced sonars provided the motivation on the investigation of Sick PLS laser scanner range measurement reliability, and line segment estimation precision. We present an approach for fitting lines straight in the lasers polar coordinate system, which enables a simple estimation of line parameter covariance. We also develop systematic error models for line parameter estimation. Finally we measure our systematic and random error models experimentally, and show, that systematic errors can be larger than random ones.