Enhancing 1D LiDAR Scanning for Accurate Stockpile Volume Estimation Within Drone-Based Mapping Systems

Abstract

Manufacturing operations such as cement plants are strongly dependent on stockpiles that serve as inputs and outputs at different stages of the production process. Estimating the stockpile volume is challenging in such environments due to high levels of dust, poor visibility, and unevenness of the stock shapes. This work proposes a simple, yet effective enhancement to drone-based mapping systems in the form of actuating low-cost 1D LiDARs via micro servo motors to increase their scanning range. The proposed solution was assessed for drone missions scanning stockpiles stored in fully confined storages under conditions similar to what would be typically found within cement plants. Simulations of the proposed aerial mapping missions were conducted in Webots simulation environment. We show that the proposed actuation of the 1D LiDAR mapping sensor can dramatically decrease volume estimation errors: absolute mean error dropped from 11% to 0.14% when mapping a stockpile with less material close to the storage walls, whereas the absolute mean error dropped from 26% to 2.6% when mapping a stockpile with more material close to the storage walls.