Tracking Targets with Known Spatial Extent Using Experimental Marine Radar Data
Jaya Shradha Fowdur, Marcus Baum, Frank J. Heymann · 2019
Endeavours to have a reliable and robust maritime traffic situation assessment are today leading to exploit the improved sensor resolution technology of radars. In such concerned tracking scenarios, multiple noisy scattered measurements arise from targets of interest's surfaces at each observation step. In this paper, we present a special version of the Multiplicative Error Model-Extended Kalman Filter* (MEM-EKF*) approach to recursively track the orientation and kinematics of targets having known physical dimensions from a maritime- and radar-based dataset. Altogether, we also introduce and thereby propose an open dataset part of a marine radar benchmark dedicated to extended target tracking to the community. The benefits and performance of the proposed approach adopted on our dataset are discussed taking into consideration the non-ideal but nonetheless prevailing effects of the quotidian marine radar such as beam spreading and false negatives. As compared to its original version, the proposed approach proved to be more computationally efficient.