Vehicle positioning method and device based on radar point cloud and storage medium

Abstract

The invention relates to a vehicle positioning method and device based on radar point cloud, and a storage medium. The method comprises the following steps of: S1, carrying out point cloud feature extraction on point cloud data collected by a radar to obtain a feature point cloud; S2, establishing a point cloud feature error model, and performing error evaluation on the feature point cloud by using the point cloud feature error model; and S3, determining the error weight of each feature point cloud based on an error evaluation result of the feature point cloud, and performing registration on the feature point cloud and map features based on the error weight so as to locate a position of a vehicle. Compared with the prior art, the vehicle positioning method and device have the advantages that the feature point cloud is obtained by extracting the point cloud features, stable structure features in the environment can be focused, the reliability of environment cognition is improved, the reliability of the point cloud features is evaluated through the point cloud feature error model, and therefore the method is used for setting and registering subsequent error weights, the registering precision is improved, and the vehicle positioning precision is improved.

Description

technical field [0001] The present invention relates to the technical field of automatic driving, in particular to a radar point cloud-based vehicle positioning method, device and storage medium. Background technique [0002] Self-driving vehicles operate in an unsupervised state. The driving brain of automatic driving can replace the driver, simulate human observation (sensor system), thinking (driving brain) and operation (planning control), and then manipulate the wheels to complete action tasks. The main functions involved in automatic driving include self-driving positioning, moving object detection, route detection, path planning, vehicle tracking control and other automatic driving functions. [0003] Accurate driving positioning is of great significance to automatic driving. Accurate vehicle positioning information is the premise of vehicle path planning and tracking control, and is the key technology to ensure automatic driving. As we all know, the accuracy of GNS...

AI Technical Summary

Problems solved by technology

If n environmental features are detected at a certain driving location, and each feature provides direction and distance constraints, the number of constraints observed in the entire environment will reach 2n, which will lead to redundant reference information and localization bias

In addition, the measurement accuracy of the same observed feature will vary with the measurement direction and distance, which will also lead to different positioning results for the same feature

At this stage, the existing automatic driving system lacks a stable evaluation of the point cloud environment around the vehicle, and adopts a homogeneous registration and positioning method for the laser point cloud. The accuracy and robustness of laser positioning are poor, which is to a certain extent Potential safety hazards for autonomous driving

Images