Method for simultaneously estimating movement and shape of target vehicle

Abstract

The invention relates to a method for simultaneously estimating the movement and shape of a target vehicle by using a preliminary distribution model of trajectory segments. The method sets a preliminary distribution model defined by a small number of representative trajectory occurrence points in real time by using trajectory segments defined in an orthogonal coordinate system and a linear correspondence between the preliminary distribution model of the trajectory segments and the center position of the target vehicle and performing many-to-one data association between trajectory segment outputs In implementation, even if the course of the target vehicle changes, the estimation performance has high reliability. A method of simultaneously estimating the movement and shape of a target vehicle using a preliminary distribution model of trajectory segments is performed by a signal processing device that processes signals transmitted from a plurality of radars mounted at a host vehicle and reflected from the target vehicle.

Description

[0001] Cross References to Related Applications [0002] This application claims the benefit of Korean Patent Application No. 10-2020-0128990 filed on October 6, 2020, the entire contents of which are incorporated herein by reference. technical field [0003] The present disclosure relates to a method of simultaneously estimating the movement and shape of a target vehicle, and more particularly, to a method using, for example, a radar tracklet segment output (radar tracklet) whose estimation performance has high reliability even when the heading of the target vehicle changes. outputs) to simultaneously estimate the movement and shape of the target vehicle. Background technique [0004] With the continuous development of self-driving vehicles, it is basically necessary to know the absolute / relative position of the surrounding environment of the self-driving vehicle (hereinafter referred to as the self-vehicle) and the self-vehicle, as well as the information of another vehicl...

AI Technical Summary

Problems solved by technology

That is, since the traditional target vehicle shape model is composed of an unmeasured number of extended measurement occurrence points (all points where the radar ray intersects the target vehicle shape model), the many-to-many relationship between the extended measurement and the target vehicle shape model The amount of computation increases geometrically during the data association process, making it impossible to implement traditional target tracking algorithms in real time based on extended measurements

[0013] Furthermore, the non-linear correspondence between the extended measurements and the position of the target vehicle leads to ambiguity when constructing the measurement equations, whereby the estimation performance may vary greatly depending on the target heading

Images