Date & Venue
Thursday, 04 October 2018, 02:00PM
Extended targets are targets which potentially generate more than one measurement per time step. They are encountered in numerous applications, spanning from vehicle tracking with automotive radar, pedestrian tracking using laser range sensors, and surface vessel tracking with marine X-band radar. The generation of multiple measurements per time step allows for the joint estimation of the target kinematics and the target extent, which generally describes size and shape. This talk presents a new method on how to perform this joint estimation through the use of random matrices. We also show simulated results of the new method, with comparisons against other extended target tracking algorithms.
Nathan Bartlett graduated from the University of Newcastle with a Bachelor of Engineering (Mechatronics) (Honours Class 1) in 2015. He is currently a PhD student studying the two related fields of simultaneous localization and mapping (SLAM), and multi-extended target tracking (METT). Utilizing a Bayesian methodology he has developed new methods to estimate both the kinematic state and geometry (extent) of extended targets in unknown environments.