Cross-Drone Binocular Coordination for Ground Moving Target Tracking in Occlusion-Rich Scenarios

Yuan Chang, Han Zhou, Xiangke Wang, Lincheng Shen, Tianjiang Hu

Abstract: How to work effectively under occlusion-rich environments remains a challenge for airborne vision-based ground target tracking, due to the natural limitation of monocular vision. Given this, a novel cross-drone binocular coordination approach, inspired by the efficient coordination of human eyes, is proposed and developed. The idea, derived from neural models of the human visual system, is to utilize distributed target measurements to overcome occlusion effects. Eventually, a binocular coordination controller is developed. It enables two distributed pan-tilt cameras to execute synergistic movements similar to human eyes. The proposed approach is able to work based on binocular or monocular vision, and hence it is practically appropriate for various environments. Both testbed experiments and field experiments are conducted for performance evaluation. Testbed experiments highlight its advantages over independent tracking in terms of accuracy while being robust to a partial perception ratio of up to 43%. Field experiments with a pair of drones further demonstrate its effectiveness in the real-world scenarios.