Implementing vision-based navigation in a hybrid 3D virtual-reality UDK-based simulator

• Main Author: Pakha, Chrisma Nasirochman
• Other Authors: Xie Lihua
• Format: Final Year Project
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics
• Online Access: http://hdl.handle.net/10356/53147
• Institution: Nanyang Technological University

This Final Year Project was taken by the author during his final year as one of the requirement to obtain degree of Bachelor Engineering. This project aims to implement vision-based navigation algorithms in a hybrid 3D simulator for multiple agent systems based on UDK. The algorithms will then be integrated into a search and exploration modular of the simulator. This Simultaneous Localization and Mapping (SLAM) problem is also known as the Kidnapped Robot Problem, a situation where a robot is carried to an arbitrary location. The simulation environment used is the Unreal Engine developed by Epic Games. Although the Unreal Engine allows simulating the robot, there are certain holes that the Unreal Engine might not be able to fill, which are the Vision System and the localization and mapping algorithm. In this project, the author focuses on integrating the SLAM algorithm to the Unreal Engine using the Robot Operating System (ROS) which runs on Ubuntu. ROS provides hardware abstraction, device drivers, libraries, message-passing and many more that might be useful for building an application for the robot in the Simulator. Up until now, two SLAM algorithms were implemented into two different robots. The GMapping was implemented on Unmanned Ground Vehicle (UGV) and the PTAM was implemented on Unmanned Aerial Vehicle (UAV). Several maps were used for testing and each SLAM algorithm produces different output. In this report the framework used and how to implement them are provided. In the end, a discussion regarding the results and improvements for future works are presented