The decentralized estimation and tracking of a mobile target performed by a group of unmanned aerial vehicles (UAVs) is studied in this work. A flocking protocol is used for maintaining a collision-free formation, while a decentralized extended Kalman filter in the information form is employed to provide an estimate of the target state. In the prediction step of the filter, we adopt and compare three different models for the target motion with increasing levels of complexity, namely, a constant velocity (CV), a constant turn (CT), and a full-state (FS) model. Software-in-the-loop (SITL) simulations are conducted in ROS/Gazebo to compare the performance of the three models. The coupling between the formation and estimation tasks is evaluated since the tracking task is affected by the outcome of the estimation process.
Comparison of Multiple Models in Decentralized Target Estimation by a UAV Swarm
2023
Aufsatz (Zeitschrift)
Elektronische Ressource
Unbekannt
Metadata by DOAJ is licensed under CC BY-SA 1.0
A cooperative guidance law with decentralized structure for target estimation
British Library Conference Proceedings | 2012
|