Formation Maintenance Control of Unmanned Aerial Vehicle Swarms based on Simulated Visual Perception

Traditional formation control methods relied on inter-agent communication are inadequate in increasingly complex electromagnetic environments. Inspired by flocking behaviors observed in biological systems, bio-inspired visual formations present a solution to overcome the limitations of conventional communication. This paper focuses on the formation maintenance control of unmanned aerial vehicle (UAV) swarms using simulated visual perception. A pairwise leader-follower three-dimensional (3D) formation control approach is proposed. The approach utilizes visual information with noise and time delays to replace the front-end vision recognition algorithm. Visual-dominant behavioral rules are designed, and an accumulated error elimination mechanism is incorporated to enhance long-term formation stability. The 3D formation maintenance algorithm was tested in software-in-the-loop simulations by changing formation sizes, flight paths, and failure conditions, proving its stability and robustness. Ablation experiments further confirmed the effectiveness of the accumulated error elimination mechanism.