Realistic simulation of unmanned aerial vehicle (UAV) executing task of large outdoor structure inspection

Unmanned aircraft algorithm testing requires expensive investments in resources. Simulators areapproaches to conduct the testing before deploying the algorithm on an actual drone. Notwithstanding many available simulators in the market, there are some trades-off using those simulators in our researc...

Full description

Saved in:
Bibliographic Details
Main Author: Eng, Yean Khai
Other Authors: Xie Lihua
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2021
Subjects:
Online Access:https://hdl.handle.net/10356/150344
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Unmanned aircraft algorithm testing requires expensive investments in resources. Simulators areapproaches to conduct the testing before deploying the algorithm on an actual drone. Notwithstanding many available simulators in the market, there are some trades-off using those simulators in our research. In this project, we aim to develop a simulator that is suitable for simulating autonomous drone mission in outdoor structural inspection scenarios. The key requirements for the simulator are (1) realistic environment providing accurate sensors output, (2) responsive physics engine forthe real-time dynamics control, and (3) flexibility to create new environments for testing purpose. The proposed simulator is intended to save researchers time and energy to set up actual testing. An advanced rendering engine from the game engine is used to render the visually realistic environment; an integrated physics engine from Gazebo is used to mimic the real-world dynamics. To examine the simulator, we have run an inspection algorithm in a self-created3 D simulated environment. The simulated aircraft completed the task accurately and acted similarly to the actual dynamic situation. Compared with the existing simulations, the key contributions of this project are: (1) simulating both visual perception sensors and Lidar point cloud sensors, (2) establishing the sensors data communication, and (3) integrating researchers’ algorithms successfully. We conclude that the simulator meets the specific requirements for our target users thereupon.