Autonomous Aerial Mapping Using a Swarm of Unmanned Aerial Vehicles

Abstract

View Video Presentation: https://doi.org/10.2514/6.2022-4062.vid Deployment of a swarm, or more generally a multi-agent robotic system, provides several advantages over a single-agent robotic system including robustness, scalability, and adaptability. In fact, when a swarm controller is designed adequately, it can provide cost-effective, fault-tolerant, and reliable solutions to many real-world applications, particularly those requiring autonomous implementation. This work demonstrates an autonomous formation control for a multi-drone system to map a stockpile within a confined space. Each employed drone is equipped with a light-weight single point 1D LiDAR mapping sensor. Given that the mapping sensor is usually the primary contributor to the system cost, we aim to demonstrate the cost-benefit of undertaking autonomous mapping missions using multiple low-cost sensors each on board of an aerial robot of the multi-agent system against using a much more expensive laser scanner on board of a single drone. To achieve this aim, a formation control strategy is developed to form the drones in a desired shape and follow a designed trajectory. The formation control was analyzed numerically and implemented within a simulation environment where an example test case of a confined space with a stockpile was considered. The stockpile was scanned with different swarm populations, and the influence of the population size on the scanning accuracy and system cost was investigated. The simulation results show that the stockpile volume was estimated with a volumetric error of -4.1% when a single drone was used and this error improved to -1.5% and -0.2% when using a multi-drone system with three and five drones, respectively. Moreover, a comparison with using the more advanced LiDAR scanners such as 2D and 3D LiDAR sensors was performed and shows that mapping with the proposed multi-drone system provides similar volume estimation results but at a lower cost.