Autonomous quadcopter for rapid underground mine mapping and exploration

Mine surveying and mapping are time-consuming and hazardous tasks that are critical to most underground mining operations. Recently, the introduction of robotics for underground mine mapping has reduced these safety risks for humans. This project aims to develop an autonomous quadcopter equipped with dedicated sensors capable of performing 3D Simultaneous Localization and Mapping (SLAM) in Global Navigation Satellite System (GNSS) denied environments, such as underground mine sites. Various localization and mapping algorithms were compared and tested, and the SpectacularAI package proved to be a computationally efficient SLAM method that can run on the quadcopter’s onboard computer, generating dense and detailed point cloud maps. Additionally, autonomous exploration and path planning methods for three dimensional spaces were studied. Appropriate algorithms were implemented and optimised to develop a software suite that ensures the platform’s autonomy in communication-limited subterranean environments. Finally, the SLAM module was tested in a cluttered indoor environment, while the autonomous exploration framework was developed and evaluated in a subterranean simulation environment. Future work will focus on deploying and testing both modules in real-world underground mine scenarios.