Probabilistic Cooperative Target Localization

This work addresses the problem of localizing a ground target observed by multiple heterogeneous unmanned vehicles. Specifically, the case of a team of unmanned aerial and ground vehicles is analyzed. Effective collaboration between unmanned aerial and ground vehicles can utilize the strengths of both platforms while mediating their individual weaknesses. In this research, a probabilistic framework is proposed to improve target localization accuracy by utilizing the epipolar geometry and inter-robot localization measurements. A virtual simulation environment is implemented to fully test the proposed methods. Hardware experiments are carried out to validate the proposed methods. Note to Practitioners-This paper was motivated by the problem of tracking ground objects of interest using multiple cooperative autonomous vehicles. Most of the existing approaches to cooperative target tracking are designed for homogeneous vehicles and neglect geometrical constraints of the multiview target tracking problem. The methodology proposed in this work incorporates heterogenous robots to cooperatively track a ground object of interest. To improve the accuracy of target localization, the epipolar geometry between multirobot views is utilized to formulate mathematical constraints. In this paper, it is analytically proved that the imposing the proposed constraints improves target localization accuracy. Experimental and simulation results demonstrate the benefits in target localization accuracy using the proposed method.