Relative state estimation enhanced collective navigation for drone swarm deprived of communication

Abstract Existing collective navigation systems for drone swarms typically rely on the communication between drones, which limits the application in specific mission scenarios and reduces the robustness against interference. To address this challenge, a communication-free collective navigation method enhanced by relative state estimation is proposed in this study. It consists of three key components: visual perception localization, relative state estimation, and swarm motion decision. First, visual sensors are employed to detect nearby drones in real time and calculate their relative positions. Second, an optimized model set adaptive interacting multiple model filtering algorithm is proposed to fuse the predicted states from the relative motion model with visual measurements to achieve continuous and high-precision relative positioning. The fused relative states are then fed back into the swarm motion decision algorithm for high-level control. Finally, the effectiveness of the proposed method is validated through numerical simulations and real-world flight experiments. The results demonstrate that the proposed relative state estimation algorithm significantly enhances the accuracy of visual relative localization and improves the performance of the swarm motion decision algorithm, enabling cohesive and collision-free navigation in communication-denied environments.