Mobile robot ego motion estimation based on a rotating 3D laser range sensor

Zadatak rada bio je razviti algoritam estimacije vlastitog gibanja mobilnog robota koji je zasnovan na mjerenjima rotirajućeg 3D laserskog senzora udaljenosti. Algoritam je implementiran u programskom jeziku C/C++ pod Robotskim operacijskim sustavom. Opisan je rad laserskog 3D rotacijskog senzora te dan opis algoritama za lasersku odometriju. Objašnjen je način izvlačenja bitnih značaji iz oblaka točaka senzora. Za izračun pomaka robota korišten je ICP algoritam koji računa transformaciju između dva oblaka točaka. Pokazana je implementacija uz pomoć biblioteke libpointmatcher te opisana njezina struktura. Na kraju su dani rezultati eksperimenta na mobilnom robotu Husky i senzoru Velodyne HDL-32E.The aim of this work was to develop a mobile robot ego motion estimation algorithm based on measurements from a rotating 3D laser sensor. The algorithm was implemented in C/C++ programming language under Robot Operating System. As a part of the introduction, the rotating 3D sensor is described and the laser odometry algorithms are presented. Also, the feature extraction from sparse point clouds is described. Mobile robot ego motion is estimated with the ICP algorithm which compares two consecutive point clouds and calculates the transformation between them. The ICP algorithm is implemented in the Libpointmatcher library so its structure is also described. At the end, the results of the experiment with the mobile robot Husky and the Velodyne HDL-32E sensor are given