A Novel Algorithm for Safe Landing of Delivery Drones in Complicated Urban Environments
The proposed algorithm utilizes path planning used by missile guidance systems to generate an optimal path for vertical landing of delivery drones
Drone delivery can expedite the supply of everyday necessities and emergency supplies. However, advancements in drone technology, such as sensor-based environment recognition and powerful processing units, do not necessarily translate into collision-free and accurate landing of drones in urban areas. In a new study, researchers now propose a novel algorithm for safe landing path planning of drones. In a real-time simulation experiment, the proposed approach outperformed EGO Planner — a state-of-the-art algorithm — in accuracy and efficiency.
In recent years, there have been rapid advancements in the study of unmanned aerial vehicles for non-military use. With the development of drone delivery technology, a wide variety of delivery services for everyday necessities as well as emergency relief efforts are now possible. Currently, drones utilize sensor-based technology with powerful graphics processing unit-enabled systems for environment recognition. However, despite these advancements, autonomous landing while safely recognizing and traversing obstacles in urban environments, remains challenging. Moreover, existing landing methods rely on the commercial global positioning system, which has an error range of several meters, making targeted landing difficult.
To overcome these challenges, deep learning techniques, such as object detection algorithms and real-time trajectory optimization using Euclidean signed distance fields, have been studied for accurate and safe drone landing. However, these methods have several limitations. They either avoid only buildings and no other obstacles in urban environments, generate candidate landing paths in advance, or are computationally expensive. If drones are to be used within civilian urban spaces, navigating the complicated landscape is indispensable.
Now, a team of researchers from Korea, led by Dr. Hanseob Lee from Digital Convergence Research Laboratory at Postal & Logistics Technology Research Center of Electronics and Telecommunications Research Institute, has addressed this research gap. In their recent study published in the ETRI Journal, they proposed a novel autonomous landing path planning algorithm for drone delivery in urban environments.
“The present work proposes a novel landing algorithm for drones in urban environments that generates an optimal landing path using the landing angle control (LAC) guidance law. We also included a rapidly exploring random tree (RRT)-based collision avoidance algorithm to avoid obstacles encountered during landing,” explains Dr. Lee.
To this end, they used the impact guidance law, an algorithm that causes a missile to collide at a specific angle at its terminal stage, to ensure that the drones always land vertically. The researchers incorporated a 3D light detection and ranging (LiDAR) sensor that detects obstacles in real time and filtered the LiDAR point cloud data using down-sampling and distance filters to remove unnecessary node points, thus reducing computational costs. In addition, the use of a receding horizon (RH) rapidly exploring random tree (RRT)-star (or RH-RRT*) algorithm ensured that the avoidance path was constantly generated throughout the landing.
The team then verified the performance of the LAC-RHRRT* algorithm using 3D simulation. They simulated a drone that operated through robot operating software with six degrees of freedom and had sensors with similar specifications as the actual drone model. In the designed simulation, the proposed algorithm outperformed state-of-the-art 3D planning technique EGO Planner in terms of accuracy and efficiency in various scenarios, such as avoiding streetlights, utility poles and wires, and trees. It generated shorter travel distances and times, and produced three times smaller landing angles.
Elaborating further, Dr. Lee says, “If commercialized, this study can ensure safe delivery of goods via drones, even within complicated urban areas. In that scenario, drones can be put to use not just for sending goods, but also for finding directions, searching for lost items, and surveying urban environments.” As per Dr. Lee, the algorithm can be put to use for applications beyond drone delivery, such as Urban Air Mobility and urban search drones.
Here’s certainly hoping that these findings expedite the practical use of drone delivery services soon!
Reference
Authors: Hanseob Lee1, Sungwook Cho2, and Hoon Jung1
Title of original paper: Real-time Collision-free Landing Path Planning for Drone Deliveries in Urban Environments
Journal: ETRI Journal
DOI: https://doi.org/10.4218/etrij.2023-0129
Affiliations:
1Digital Convergence Research Laboratory, Postal & Logistics Technology Research Center, Electronics and Telecommunications Research Institute
2Department of Aeronautical and Mechanical Engineering, Cheongju University
Your Press Release Source
Electronics and Telecommunications Research Institute
About Electronics and Telecommunications Research Institute (ETRI)
ETRI is a government-funded research institute that was established in 1976 in Daedeok Science Town in Daejeon, Republic of Korea. ETRI makes contributions to the nation’s economic and social development through research, development, and distribution of industrial core technologies in the fields of information, communications, electronics, broadcasting, and convergence technologies. Its vision is “Technology Pioneer Making Happy Future through Digital Innovation”. To learn more about ETRI and its exciting opportunities for researchers and students alike, please visit https://www.etri.re.kr/eng/main/main.etri.
About Dr. Hanseob Lee
Hanseob Lee is a researcher at the Digital Convergence Research Laboratory at the Postal & Logistics Technology Research Center of Electronics and Telecommunications Research Institute (ETRI), Daejeon, Republic of Korea. His group is developing autonomous delivery robots and drones. His research interests include autonomous applications in aerial robotics, continuous reinforcement learning, and vision-based tracking. He received his MS and PhD degrees in aerospace engineering from Korea Advanced Institute of Science and Technology, Republic of Korea, in 2017 and 2021, respectively. He has published more than ten papers catering to topics like robotics, automation, flight dynamics, reinforcement learning, and UAVs.
