The proposed swarm framework consists of several interconnected modules:
Model and State Estimation of Surroundings: A bank of Linear Kalman Filters is used to model and estimate the state of observable UAVs (oUAVs) in the swarm. This provides a reliable neighborhood model for the high-level control.
Flocking Control: A state feedback control law is designed to specify the desired group velocity and stabilize each UAV in an unambiguous position within the swarm formation. This approach allows for significantly higher swarm velocities compared to standard reactive flocking algorithms.
Enhanced Multi-Robot State Estimation (MRSE): The onboard state estimation is enhanced by adaptively fusing Visual Inertial Odometry (VIO) with the cooperative state estimation. This improves the reliability of the purely onboard localization in feature-poor environments.
Velocity Estimation of Observable UAVs: To decrease the dependence on unreliable communication networks, a method is introduced to estimate the velocities of neighboring UAVs based on the observed swarming behavior, without explicit communication.
The proposed framework was extensively validated through complex real-world experiments, demonstrating its capability to achieve high group velocities up to the physical limits of the hardware, while maintaining swarm cohesion without reliance on GNSS and communication.
To Another Language
from source content
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Jiri Horyna,... lúc arxiv.org 04-30-2024
https://arxiv.org/pdf/2404.18729.pdfYêu cầu sâu hơn