Vehicle-mounted radar data association method and device

Abstract

The invention provides a vehicle-mounted radar data association method and device. The method comprises the steps of setting two sets of data association thresholds for each track prediction point, namely, a first data association threshold and a second data association threshold; for each track prediction point, determining whether the measurement value is a candidate measurement value of the track prediction point or not through the first data association threshold and the cumulative number of times that the measurement value falls into the data association thresholds of other track prediction points; and weighting all candidate measurement values of the track prediction point, and taking the weighted candidate measurement values as final association data of the track prediction point to be used for updating the track, so that the phenomenon of misassociation is reduced. The measurement value which falls into the second data association threshold and does not fall into the first data association threshold is labeled, so that the measurement value cannot be used as a new track start, false alarm of the same target caused by large measurement value distribution is inhibited, and track continuity and stability are improved.

Description

technical field [0001] The present invention relates to the technical field of vehicle radar data processing, and more specifically, relates to a vehicle radar data association method and device. Background technique [0002] In a single-target clutter-free environment, there is only one target's measurement value within the detection range of the radar. At this time, there is no data association link in data processing, and only the tracking problem is involved. In the case of multiple targets and complex scenes on the actual road, there are many target measurement values ​​within the detection range of the radar, and the number of measurement values ​​is uncertain. At this time, data association problems will be involved. The data association problem is to establish the correct relationship between the radar measurement value and the existing tracking track at a certain moment, so as to determine whether these measurement values ​​come from the same target, and finally rea...

AI Technical Summary

Problems solved by technology

Compared with other algorithms, the joint probabilistic data association algorithm has stronger anti-interference ability for the environment containing a lot of clutter, and is more suitable for the environment with dense target clutter, but when the target is too dense and concentrated and contains a lot of clutter When , the joint probabilistic data association algorithm often has the problem of combinatorial explosion, that is, the amount of computation increases exponentially with the linear increase of the target, which eventually leads to a sharp decline in the real-time performance and stability of the tracking algorithm.

[0010] Based on the basic principles of the above-mentioned probabilistic data association algorithm and joint probabilistic data association algorithm, for figure 2 In the scenario shown, the point p is predicted according to the calculated track 1 The data association threshold S 1 and track prediction point p 2 The data association threshold S 2 , although the track prediction point p 1 and p 2 can achieve data association, but the measured value z 3 Belongs to track prediction point p 1 and p 2 shared, measured value z 3 Does not belong to the track prediction point p at all 1 , and finally enters the data association threshold S 1 The measured value of z 3 and z 4 Weighted as track prediction point p 1 associated value; but obviously z 3 The intervention of will affect the track prediction point p 1 processing accuracy

What is more serious is that originally belonged to the track prediction point p 1 The measured value of z 2 , did not enter the track prediction point p at all 1 The data association threshold S 1 , so that the measured value z 2 will start as an independent track, so it will lead to track prediction point p 1 false alarm nearby

Images