Probabilistic Path Planning for Cooperative Target Tracking in Urban Area Using UAVs

In this paper, we propose a probabilistic path planning algorithm to track multiple ground moving targets using unmanned aerial vehicles. The algorithm takes into account the occlusion of drone sensors by urban buildings and uses a dynamic occupancy grid network and Bayesian filter model to represent and update the location distribution of the target. Using the current position of the UAV and the predicted distribution of the target positions, the algorithm plans the path of a single UAV to achieve the goal of minimizing the possible distribution area of the targets over a period of time. The tracking targets of multiple drones are allocated using a clustering algorithm based on graph segmentation and an auction algorithm, thereby generating path planning results for multiple drones to track multiple city targets. Monte Carlo experimental results based on random scene generation show that the algorithm can solve the target tracking problem in urban environments.