Iterative Model Matching for 3D Localization of a Mobile Robot
Description
In this study, an algorithm that iteratively matches the following models obtained from point clouds is proposed for the 3D localization of a mobile robot. Edges derived from point clouds of each scan sampling with a Kinect. After defining a range of interest, the algorithm creates a model by fitting the classified points according to their slope and separation distance. The obtained models are compared with the consecutive closest one of the previous scans in an iterative manner. A transformation matrix is derived from this comparison that calculates displacement and change of orientation of the robot.