A Method for Optimizing LiDAR Localization Using Radius Search

Abstract

The invention belongs to the technical field of road network navigation, and provides a method for searching and optimizing laser radar positioning by a radius. A multi-thread programming scheme is employed, input GPS coordinate is used as a center in an another thread, N pose parameters are generated by a method F within a range of the radius being R and height being H, and matching is constantlytried and ended until set matching accuracy S is met. The method is used for positioning a position, needed to be positioned, of a vehicle, local road distribution and vehicle arrival region information are inquired in database by GPS rough coordinate, the post which the vehicle cannot reach is filtered when forecast post is generated, the number of the forecast post can be greatly reduced, so that matching accuracy and speed are improved.

Description

technical field [0001] The invention relates to radio navigation in the field of automatic driving, in particular to a method for optimizing laser radar positioning by using radius search. Background technique [0002] Vehicle positioning is a very important link in the automatic driving system, which is used to determine the orientation and position of the car in the high-precision map. There are many ways of vehicle positioning, among which using lidar point cloud for matching is a relatively common way, NDT-normal distribution transform (Normal Distributions Transform, NDT), hereinafter referred to as ndt, is a common lidar point cloud matching algorithm , the principle of the algorithm is to convert the target point cloud and high-precision point cloud map information generated by lidar into a normal distribution of multi-dimensional variables. By calculating the distribution probability, if the two have a high matching degree, the scanned point cloud and map position p...

AI Technical Summary

Problems solved by technology

There are many ways of vehicle positioning, among which using lidar point cloud for matching is a relatively common way, NDT-normal distribution transform (Normal Distributions Transform, NDT), hereinafter referred to as ndt, is a common lidar point cloud matching algorithm , the principle of the algorithm is to convert the target point cloud and high-precision point cloud map information generated by lidar into a normal distribution of multi-dimensional variables. By calculating the distribution probability, if the two have a high matching degree, the scanned point cloud and map position pose matching, although ndt has the advantages of high efficiency and offline calculation, but due to the limitation of the algorithm, for the same lidar point cloud, there will be multiple possible matching pose points in the high-precision map. Accuracy, the rough pose of the lidar point cloud needs to be passed as input to ndt. In the current common practice, this pose is given by GPS positioning, but due to GPS accuracy and drift problems, it will lead to the estimation given each time The postures are not the same, which may lead to the possibility that the required matching accuracy cannot be achieved for a long time, and even the vehicle positioning may be lost.

Images