Impact of Communication Delays on Multi-Drone Cooperative Navigation

This paper presents a decentralized cooperative navigation algorithm aimed at preventing divergence in state estimation by accurately modeling cross-covariance terms and compensating for communication delays occurring during data transmission among platforms, referred to as Operating Agents (OAs). Each agent independently updates its own state using non-cooperative measurements. When cooperative measurements become available, a designated Computing Agent (CA), selected iteratively from within the swarm, gathers the shared information, performs the cooperative update, and sends the updated state and covariance to the other agents. A detailed analysis of potential sources of communication delays and their effect on filter performance is conducted. The core innovation of the proposed approach lies in the introduction of a delay compensation strategy, along with a performance assessment comparing compensated and uncompensated filter configurations. The algorithm is evaluated in two simulated scenarios involving three and five fixed-wing UAVs, leveraging GNSS, IMU, magnetometer, and inter-agent ranging measurements. Additional ranging measurements are provided with respect to mobile landmarks, referred to as Supporting Agents (SAs) introduced to address potential GNSS-challenging areas where filter performance may degrade. Results demonstrate that delay compensation improves state estimation performance, particularly when the standard deviation of the delay is large and the number of OAs is limited. Finally, experimental validation is proposed using a three-drone swarm setup, each equipped with an IMU, a magnetometer, a GNSS receiver, and an RF ranging device, with sensor acquisition managed through a custom ROS-based network architecture.