LTGPv2: Rethinking local track geometry for Track-to-Track Association

Abstract

Track-to-track association (T2TA) is an essential part in situational awareness of advanced driving assistant systems. The accuracy of track-to-track association methods may degrade by missed detections and measurement bias. Although the local track geometry preservation algorithm has been proposed to improve the performance of T2TA, it may be affected as the target detection decreases. In this study, we proposed second version of the local track geometry preservation algorithm, called LTGPv2, which introduces another local track structure descriptor for T2TA. The local tracks of one sensor are represented by Gaussian mixture model (GMM) centroids, and are fitted to the corresponding local tracks of the other sensor. The T2TA problem is formulated as a maximum likelihood estimation problem with two local track geometry constraints to avoid the degradation of T2TA performance caused by missing detection. Then, an expectation–maximization (EM) algorithm is applied to solve it. Simulation results demonstrate that LTGPv2 obtain better performance than the state-of-the-art methods.