Master Thesis Omer Rajput 3 D Modelling , Feature Extraction and Segmentation of Modern Warehouse Environments using Laserscanners September 10 , 2014

Current logistic systems are the bottleneck in the factory automation. The automation in the logistics is only limited to the point of providing manual maps. The creation of these maps can take months for a moderately sized warehouse. The current logistic system can benefit from automatic generation of maps in 3D with semantic information associated with it. This work provides a methodology to generate 3D maps with semantic information automatically from the data acquired through an special arrangement of SICK 2D laserscanners. A series of algorithms are developed in this work to solve the complicated problem of identifying natural landmarks by developing a variety of algorithms suited for various landmarks present in a modern warehouse. This work successfully handles the dataset of ∼800 million points (∼37GiB) representing a modern warehouse, and provides a semantic representation of the warehouse by means of the segmented objects like floor, ceiling, pillars, poles, racks, etc. This work provides a solution that improves the detection rate and the run-time of the algorithms due to the development of novel techniques.