RADAR vs LIDAR for obstacle detection and collision avoidance

Tiago Filipe Félix de Andrade

Key information

Authors:

Tiago Filipe Félix de Andrade (Tiago Filipe Félix de Andrade)

Supervisors:

Alexandra Bento Moutinho (Alexandra Moutinho)

Published in

10/12/2020

Abstract

The field of autonomous vehicles is a dynamic and ever-growing field of research which has seen a rise in popularity in recent years. Research in this area has mostly shifted towards development of increasingly accurate and versatile sensing systems to be integrated in robust autonomous navigation algorithms, culminating in highly autonomous vehicles. Diversity in the field of sensing is significant, with a high number of developed sensing methods to acquire a variety of environment data used in autonomous navigation. This thesis focuses on two of the most commonly used sensors, LIDAR and Doppler RADAR, exploring the technology behind these devices. The main goal of this work is the development of an obstacle detection and collision avoidance system incorporating these sensors and its implementation in a small-scale robot capable of autonomously navigating unknown environments to reach a single or multiple waypoints. A physical prototype is developed as well as its virtual counterpart to simulate its behaviour in a controlled, risk-free environment intended for prior testing. Modularity is a significant factor in this work, with the intent of easing the integration of the prototype in any Wi-Fi network. The capabilities of the ROS framework are explored with its integration in the assembly. Finally, the sensing system is tested in a set of trials, evaluating the performance of the standalone sensors as well as its coupling (through sensor fusion) in static and dynamic environments.