Autonomous system for 3D mapping of indoor environments

Abstract

This document describes the software and hardware specifications for the project titled ‘Autonomous 3D mapping of indoor environments’. This project aims to create a software and hardware platform that will be able to generate accurate 3D models of building interiors using inexpensive hardware and state of the art techniques. This system will allow users to quickly visualize real world environments for purposes such as simulations, interior design and Virtual reality applications. This project proposes a solution consisting of a Microsoft Kinect for data acquisition and an Nvidia Jetson tx2 development kit for processing. RGBD data is collected by scanning any physical space using the kinect. Afterwards, data is segmented using region growing and then Random Sample Consensus is applied to the resulting regions.Next, vertices and edges of the intersecting planes are found using mathematical techniques. Finally, all the planes in the image are generated using their mathematical equations to create a 3D model of the scanned area. The greatest challenge for this project was getting the hardware components to work together and locate compatible drivers and libraries. This was the most time consuming and demanding part of the project. The logical next step is to refine the prototype and turn it into a marketable solution for the local market. However, there is a lot of work to be done before this product reaches a market ready state.