Method and system for matching ground elements between multiple batches of relative maps

Abstract

The invention relates to a matching method and system for ground elements between multiple batches of relative maps, and the matching method comprises the steps: selecting map frames of the relative maps and key points of the map frames, segmenting each map frame by taking the key points as the center, and outputting the frame range of the map frames; obtaining the types and the corresponding number of ground objects with reliable positions in the ground object objects contained in the frame range, and determining key frames in the map frame according to the types and the corresponding number of the contained ground objects; comparing the ground object to match the neighbor key frames, and outputting an association relationship of the neighbor key frames; after the global association result is adjusted, outputting a ground feature association result; matching is carried out based on semantic elements, data error tolerance is higher, an actual production environment is better met, and the method is easy to debug, convenient to control, good in interpretation, high in robustness, capable of directly expressing data features and suitable for being used in a real environment of large-scale data.

Description

technical field [0001] The invention relates to the field of high-precision map production, in particular to a method and system for matching ground elements between multiple batches of relative maps. Background technique [0002] In the field of high-precision map production, it often involves the fusion of multiple batches of relative maps. The relative map refers to the local map constructed by the camera and IMU (inertial measurement unit) when the bicycle is moving, and is registered to the absolute position through the low-precision GPS reference. The absolute accuracy of the ground object is about 1-10m. The relative map of a single batch refers to the result of a single run of a bicycle. The multi-batch relative map refers to the result of multi-vehicle running for multiple times, which is actually multiple single-batch relative maps. [0003] The matching of ground objects between multiple batches of relative maps is to estimate which objects of different batches ...

AI Technical Summary

Problems solved by technology

[0004] The absolute accuracy of relative maps based on low-precision GPS is generally within 10 meters. In the process of multi-batch relative map fusion, the matching process of ground elements is the most critical. When the ground elements of different batches of maps can be correctly matched and correlated, The subsequent optimization can only be continued, and the matching method of ground elements is often difficult to obtain reliable and stable results due to the environment of the collected data, changes in data quality and accuracy

[0005] Conventional different batches of ground feature matching methods include image semantic descriptor matching in SLAM, but in actual situations, the descriptor extraction methods of these methods are mostly obtained from experience, and the results cannot always match the actual environment

Images