TY - GEN
T1 - An Efficient Formation Control mechanism for Multi-UAV Navigation in Remote Surveillance
AU - Raja, Gunasekaran
AU - Baskar, Yashvandh
AU - Dhanasekaran, Priyanka
AU - Nawaz, Raheel
AU - Yu, Keping
N1 - Funding Information:
Gunasekaran Raja, Yashvandh Baskar, Priyanka Dhanasekaran gratefully acknowledge the support from NGNLab, Department of Computer Technology, Anna University, Chennai, India.
Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - Multiple Unmanned Aerial Vehicles (UAVs) have a greater potential to be widely used in civil and military applications. Swarm of UAVs can be deployed in a multitude of 24/7 security and surveillance. The network management and pattern formation are crucial for multi-UAV formation control mechanisms while cautiously navigating the surveillance areas. A Deep Reinforcement Learning (DRL) based Formation Flight Control for Navigation (FFCN) is used to efficiently build the UAV swarm, which decreases networking load by minimizing communication and processing involved in pattern formation. Moreover, through the leader-follower navigation, the network management of the swarm is substantially simplified. The leader-follower approach in FFCN is efficient for multi-UAV as the navigation system needs to find only the leader's trajectory. However, the failure of the leader due to actuator faults decreases the efficiency of the system. The proposed FFCN addresses the above by including a fault-tolerance mechanism, thus improving the system's reliability. Simulation results show that the FFCN model achieves faster convergence in less time with a lower collision rate. The model's usage reduced the collision rate to 3.4% in successful formation without colliding with other UAVs.
AB - Multiple Unmanned Aerial Vehicles (UAVs) have a greater potential to be widely used in civil and military applications. Swarm of UAVs can be deployed in a multitude of 24/7 security and surveillance. The network management and pattern formation are crucial for multi-UAV formation control mechanisms while cautiously navigating the surveillance areas. A Deep Reinforcement Learning (DRL) based Formation Flight Control for Navigation (FFCN) is used to efficiently build the UAV swarm, which decreases networking load by minimizing communication and processing involved in pattern formation. Moreover, through the leader-follower navigation, the network management of the swarm is substantially simplified. The leader-follower approach in FFCN is efficient for multi-UAV as the navigation system needs to find only the leader's trajectory. However, the failure of the leader due to actuator faults decreases the efficiency of the system. The proposed FFCN addresses the above by including a fault-tolerance mechanism, thus improving the system's reliability. Simulation results show that the FFCN model achieves faster convergence in less time with a lower collision rate. The model's usage reduced the collision rate to 3.4% in successful formation without colliding with other UAVs.
KW - Collision Avoidance
KW - Fault-Tolerance and Deep Q Network (DQN)
KW - Pattern Formation
KW - Remote Surveillance
KW - Unmanned Aerial Vehicles (UAVs)
UR - http://www.scopus.com/inward/record.url?scp=85126091835&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85126091835&partnerID=8YFLogxK
U2 - 10.1109/GCWkshps52748.2021.9682094
DO - 10.1109/GCWkshps52748.2021.9682094
M3 - Conference contribution
AN - SCOPUS:85126091835
T3 - 2021 IEEE Globecom Workshops, GC Wkshps 2021 - Proceedings
BT - 2021 IEEE Globecom Workshops, GC Wkshps 2021 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2021 IEEE Globecom Workshops, GC Wkshps 2021
Y2 - 7 December 2021 through 11 December 2021
ER -