DESIGN AND DEVELOPMENT OF AUTONOMOUS UAVS FOR STOCKPILE VOLUME ESTIMATION IN OPEN-CAST MINES

Abstract

The use of Unmanned Aerial Vehicles (UAVs) in stockpile and dump volume estimation in open cast mines has become increasingly prevalent due to their efficiency and accuracy. Traditional methods of volume estimation, such as manual measurement, GNSS-RTK surveys, and terrestrial laser scanning, are labour-intensive, time-consuming, and expensive. UAV technology offers a dynamic and innovative approach, capable of capturing high-resolution data over extensive areas within a short time frame. This dissertation utilizes LiDAR point clouds and photogrammetric data collected through UAVs to address these challenges. The proposed methodology offers various advantages, including improved accuracy in volume calculations compared to conventional methods, reduced labour requirements, decreased time consumption, and cost-effectiveness. Various case studies are conducted to determine and verify the effectiveness and accuracy of these methods for estimating stockpile volumes. Three quadcopters with different payload capacities are designed and built for learning and training, carrying survey-grade cameras and airborne LiDAR sensors, respectively. The design and analysis of the quadcopter frames are conducted to ensure reliability and performance. Flight tests, parameter tuning, and vibration analysis are performed to make the quadcopters stable, reliable, and responsive under harsh conditions in open cast mines. Ground Control Points (GCP) or RTK-enabled drones are essential for achieving survey-grade accuracy in volume estimation, which makes the initial cost of equipment acquisition expensive. A marker group-based methodology is proposed and verified to determine volumes without the need for a GNSS-RTK system or RTK-enabled drone. The proposed methodology yields accurate results comparable to those obtained using GCPs, offering a similar level of accuracy.