Random Finite Set based Safe Landing Zone detection and tracking

This paper addresses a robust Safe Landing Zone (SLZ) perception method for the autonomous landing of UAV. First, a detection method for identifying SLZs is presented, which makes use of the RANSAC plane fitting algorithm and several constraints. Meanwhile, the inherent uncertainties in the navigational and perception sensors of UAV make SLZ detection and tracking difficult. Therefore, in this paper, the state of SLZs and UAV is modeled as a Random Finite Set, and the tracking process is implemented using the Gaussian Mixture Probability Hypothesis Density (GM-PHD) filter. The performance of the proposed systems is demonstrated in the GAZEBO simulation environment’s volcano map.