Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Calibration of a multi-beam laser system by using a TLS-generated reference

: Gordon, Marvin; Meidow, Jochen

Fulltext urn:nbn:de:0011-n-2667653 (2.6 MByte PDF)
MD5 Fingerprint: 63945a651014a9ae4e09573f40bd8701
Created on: 19.11.2013

Scaioni, M. ; International Society for Photogrammetry and Remote Sensing -ISPRS-:
ISPRS Workshop Laser Scanning 2013 : 11 – 13 November 2013, Antalya, Turkey
Istanbul: ISPRS, 2013 (ISPRS Annals II-5/W2)
International Workshop Laser Scanning <2013, Antalya>
Conference Paper, Electronic Publication
Fraunhofer IOSB ()
LIDAR; calibration; estimation; TLS; reference data; mapping

Rotating multi-beam LIDARs mounted on moving platforms have become very successful for many applications such as autonomous navigation, obstacle avoidance or mobile mapping. To obtain accurate point coordinates, a precise calibration of such a LIDAR system is required. For the determination of the corresponding parameters we propose a calibration scheme which exploits the information of 3D reference point clouds captured by a terrestrial laser scanning (TLS) device. It is assumed that the accuracy of this point clouds is considerably higher than that from the multi-beam LIDAR and that the data represent faces of man-made objects at different distances. After extracting planes in the reference data sets, the point-plane-incidences of the measured points and the reference planes are used to formulate the implicit constraints. We inspect the Velodyne HDL-64E S2 system as the best-known representative for this kind of sensor system. The usability and feasibility of the calibration procedure is demonstrated with real data sets representing building faces (walls, roof planes and ground). Beside the improvement of the point accuracy by considering the calibration results, we test the significance of the parameters related to the sensor model and consider the uncertainty of measurements w.r.t. the measured distances. The Velodyne returns two kinds of measurements—distances and encoder angles. To account for this, we perform a variance component estimation to obtain realistic standard deviations for the observations.