TY - JOUR
T1 - Visual Collaboration Leader-Follower UAV-Formation for Indoor Exploration
AU - Evangeliou, Nikolaos
AU - Chaikalis, Dimitris
AU - Tsoukalas, Athanasios
AU - Tzes, Anthony
N1 - Funding Information:
This work was partially performed at the KINESIS CTP facility, New York University Abu Dhabi, Abu Dhabi, 129188, United Arab Emirates.
Publisher Copyright:
Copyright © 2022 Evangeliou, Chaikalis, Tsoukalas and Tzes.
PY - 2022/1/4
Y1 - 2022/1/4
N2 - UAVs operating in a leader-follower formation demand the knowledge of the relative pose between the collaborating members. This necessitates the RF-communication of this information which increases the communication latency and can easily result in lost data packets. In this work, rather than relying on this autopilot data exchange, a visual scheme using passive markers is presented. Each formation-member carries passive markers in a RhOct configuration. These markers are visually detected and the relative pose of the members is on-board determined, thus eliminating the need for RF-communication. A reference path is then evaluated for each follower that tracks the leader and maintains a constant distance between the formation-members. Experimental studies show a mean position detection error (5 × 5 × 10cm) or less than 0.0031% of the available workspace [0.5 up to 5m, 50.43° × 38.75° Field of View (FoV)]. The efficiency of the suggested scheme against varying delays are examined in these studies, where it is shown that a delay up to 1.25s can be tolerated for the follower to track the leader as long as the latter one remains within its FoV.
AB - UAVs operating in a leader-follower formation demand the knowledge of the relative pose between the collaborating members. This necessitates the RF-communication of this information which increases the communication latency and can easily result in lost data packets. In this work, rather than relying on this autopilot data exchange, a visual scheme using passive markers is presented. Each formation-member carries passive markers in a RhOct configuration. These markers are visually detected and the relative pose of the members is on-board determined, thus eliminating the need for RF-communication. A reference path is then evaluated for each follower that tracks the leader and maintains a constant distance between the formation-members. Experimental studies show a mean position detection error (5 × 5 × 10cm) or less than 0.0031% of the available workspace [0.5 up to 5m, 50.43° × 38.75° Field of View (FoV)]. The efficiency of the suggested scheme against varying delays are examined in these studies, where it is shown that a delay up to 1.25s can be tolerated for the follower to track the leader as long as the latter one remains within its FoV.
KW - drone
KW - leader follower
KW - relative localization
KW - swarm
KW - UAV
UR - http://www.scopus.com/inward/record.url?scp=85123178188&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85123178188&partnerID=8YFLogxK
U2 - 10.3389/frobt.2021.777535
DO - 10.3389/frobt.2021.777535
M3 - Article
AN - SCOPUS:85123178188
VL - 8
JO - Frontiers in Robotics and AI
JF - Frontiers in Robotics and AI
SN - 2296-9144
M1 - 777535
ER -