Centralized UAV Swarm Formation Estimation with Relative Bearing Measurements and Unreliable GPS

John Akagi, Randall S. Christensen and Matthew W. Harris

Peer Reviewed

Abstract:	Unmanned Aerial Vehicles (UAVs) are used in a variety of tasks from package delivery to infrastructure inspection. In these situations UAVs typically rely on GPS signals to measure their current position. When operating in certain locations, like within urban canyons or where GPS signals are being disrupted, UAVs need alternate means to estimate their position and avoid collision. Current techniques include using LIDAR or camera measurements to detect nearby UAV and other objects. This paper presents a centralized Extended Kalman filter where the states of a UAV swarm are estimated using line of sight measurements from a camera. A single UAV is assumed to be able to receive sporadic GPS measurements but that information is able to propagate through the swarm via the relative measurements. Other research has shown that camera measurements can be used in a leader-follower situation but the contribution of this work is applying camera measurements to a swarm setting. The results show that this method does allow for the GPS denied UAVs to maintain a reasonable estimate of their state. Additionally, the availability of the line of sight measurements is adjusted and the change in state estimation accuracy is seen to depend primarily on the formation geometry. These results allow designers to gauge what measurement frequency is necessary to support a desired level of state estimation.
Published in:	2020 IEEE/ION Position, Location and Navigation Symposium (PLANS) April 20 - 23, 2020 Hilton Portland Downtown Portland, Oregon
Pages:	383 - 391
Cite this article:	Akagi, John, Christensen, Randall S., Harris, Matthew W., "Centralized UAV Swarm Formation Estimation with Relative Bearing Measurements and Unreliable GPS," 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), Portland, Oregon, April 2020, pp. 383-391.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In