Likelihood Adjustment Based on Informativeness of Observation for Extended Object Tracking
Koichiro Suzuki, Norikazu Ikoma, Mitsutoshi Morinaga, Chiharu Yamano · 2019
In this paper, we consider the problem of extended object tracking when a few measurements are available. For automotive safety, it is important to estimate the state of a vehicle, that is, position, velocity, and vehicle's extent, using multiple measurements from unique objects. Many observation models of extended object tracking for the vehicle tracking problem have been proposed, but they have paid less attention to a number of measurements that can degrade the likelihood of target. We propose a novel likelihood adjustment method based on the informativeness of observation, which corresponds to the number of observed points and their formation, in a context of extended object tracking. Numerical simulations with a simplified but essential model figure performance of the proposed method by comparing with conventional methods.