Developing an autonomous, task-distributed drone network for the mapping of remote and isolated targets


Developing an autonomous, task-distributed drone network for the mapping of remote and isolated targets

Lorenz, S.; Booysen, R.; Madriz Diaz, Y. C.; Thiele, S. T.; Kirsch, M.; Gloaguen, R.

Uncrewed aerial vehicles (UAVs), also referred to as drones, have become a major developing branch in the field of autonomous vehicles. Lightweight, flexible, and inexpensive, UAVs can offer individual solutions for a wide range of applications. Important assets are the fast turnaround times and high customizability of UAV platforms and their respective payloads. This targeted and adapted surveying allows us to map chemical and physical properties of complex or even inaccessible terrains. Regulatory and technical barriers, however, limit the product of take-off weight and endurance for civil and research use. Common compromises are light-weight systems with high ground coverage and small payloads (e.g., small, fixed-wing drones), and heavy-duty UAV (e.g., multi-copters) with shorter flight times. The latter in turn provide the opportunity to deploy heavier, highly technological equipment to gather more information on the depicted scene.
This trade-off causes a dilemma, in particular for drone-borne material mapping with spectral imaging sensors. Light-weight systems can achieve sufficient aerial coverage within a reasonable time, however, light-weight cameras are mostly limited to uncooled systems covering the visible and near-infrared range of the electromagnetic spectrum. Such sensors allow characterization only for a limited number of materials with often low confidence. The lack of subsurface information acquired with these sensors further limits the provided data value, especially in regions with extensive vegetation or soil coverage.
Systems analyzing subsurface geophysical properties or providing enhanced spectroscopic information (e.g., by extending the detection range towards longer wavelengths) are often heavier and/or require adapted drone design and flight planning. Using such systems to cover a full prospect area at the required detail is tedious. Slow flight speeds, repeated battery changes, and a tremendous amount of data to process cause often intolerable delays. Short turnaround times, however, are key in the respective application fields, as either environmental conditions or the mission itself may offer a limited time-window for data acquisition and initial result delivery. This is a major hurdle for many potential UAV applications such as greenfield mineral exploration, search & rescue, or leak/pollution detection, where targets of interest are often remote, small-scaled and of unknown exact location.
We present an innovative concept capable of performing rapid and reliable target characterization via a domain approach. The core idea is the development of a task-distributed drone network, combining the strengths of light-weight and heavy-duty systems. As high-detail data is only acquired where it matters, long flight-times and large volumes of superfluous data can be avoided from the start. This also reduces processing time, computational requirements, as well as the impact of the survey on the environment. As a first step, we demonstrate the challenges and opportunities provided by such multi-modal, drone-based mapping in the framework of mineral exploration. In several case studies, we also showcase the added value of integrating surface (spectral imaging) and subsurface (geophysical) data for better target characterization and give an outlook on autonomous and multi-drone data acquisition for a targeted and more efficient characterization.

  • Poster (Online presentation)
    Workshop on Hyperspectral Image and Signal Processing : Evolution in Remote Sensing (WHISPERS), 13.-16.09.2022, Roma, Italia

Permalink: https://www.hzdr.de/publications/Publ-35848
Publ.-Id: 35848