4 June 2019 in IEEE Transactions on Aerospace and Electronic Systems
This paper presents a Poisson multi-Bernoulli mixture (PMBM) conjugate prior for multiple extended object filtering. A Poisson point process is used to describe the existence of yet undetected targets, while a multi-Bernoulli mixture describes the distribution of the targets that have been detected. The prediction and update equations are presented for the standard transition density and measurement likelihood. Both the prediction and the update preserve the PMBM form of the density, and in this sense the PMBM density is a conjugate prior. However, the unknown data associations lead to an intractably large number of terms in the PMBM density, and approximations are necessary for tractability. A gamma Gaussian inverse Wishart implementation is presented, along with methods to handle the data association problem. A simulation study shows that the extended target PMBM filter performs well in comparison to the extended target $\delta$ -GLMB and LMB filters. An experiment with Lidar data illustrates the benefit of tracking both detected and undetected targets.
29 December 2017 in IEEE Transactions on Intelligent Vehicles
Environment perception is a key enabling technology in autonomous vehicles, and multiple object tracking is an important part of this. The use of high resolution sensors, such as automotive radar and lidar, leads to the extended object tracking problem, with multiple detections per tracked object. For computationally feasible multiple extended object tracking, the data association problem must be handled. Previous work has relied on a two step approach, using clustering algorithms, together with assignment algorithms, to achieve this. In this paper, we show that it is possible to handle the data association in a single step that works directly on the desired likelihood function. Single step data association is beneficial, because it enables better use of the measurement model and the predicted multi-object density. For single step data association, we use algorithms based on stochastic sampling, and integrate them into a Poisson Multi-Bernoulli Mixture filter. In a simulation study, and in an experiment with Velodyne data acquired in an urban environment, four sampling algorithms are compared to clustering and assignment. The results from the simulations and the experiment show that single-step likelihood-based data association achieves better performance than two-step clustering and assignment data association does.