Global Position Orientation Correction for HD Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the creation of high definition maps for autonomous vehicles, existing methods face challenges with local rotation offset issues during point cloud registration, leading to overlapping data and reduced map accuracy, particularly when convergence is not achieved in the solution process using GNSS and IMU data.
Innovation Solution
A method that corrects global position and orientation information by determining pairs of trajectory points, obtaining relative position and orientation estimation and measurement information, and calculating a local rotation offset using a mileage measurement device to improve map accuracy and reduce overlapping phenomena.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If global position and orientation information is obtained using GNSS and IMU data, then the overall positioning framework is established, but local rotation offset issues occur leading to reduced map accuracy
Solution Approach 1:
The patent segments the positioning problem into global positioning (using GNSS/IMU) and local rotation correction (using trajectory point pairs). By dividing the overall positioning task into coarse global positioning and fine local rotation adjustment, the system maintains the reliability of the global framework while improving local map accuracy through separate correction of rotation offsets.
Solution Approach 2:
The patent introduces trajectory point pairs as an intermediary element to bridge global positioning data and local map registration. These trajectory point pairs serve as a mediator that connects the global coordinate system with local point cloud data, enabling the detection and correction of rotation offsets that occur in the global positioning process.
2Productivity
If point cloud registration is performed based on global position and orientation information, then the map construction process is completed, but overlapping data phenomena occur reducing registration quality
Solution Approach 1:
The patent implements a feedback mechanism where trajectory point pairs provide measurement information about actual relative positions and orientations. This feedback is used to calculate rotation offsets and correct the global position and orientation information before point cloud registration, ensuring that the registration process uses corrected data that prevents overlapping phenomena.
Solution Approach 2:
The patent performs preliminary correction of global position and orientation information using trajectory point pairs before the actual point cloud registration process. By pre-calculating and applying rotation offset corrections to the global coordinate data beforehand, the system prevents registration errors and overlapping phenomena from occurring in the first place.
3Manufacturing precision
If local rotation offset correction is implemented using trajectory point pairs, then map accuracy is improved, but calculation complexity increases
Solution Approach 1:
The patent applies local quality by focusing correction efforts specifically on local rotation offsets at trajectory points rather than attempting to correct the entire global coordinate system. By identifying and correcting only the local rotation issues where they occur (at specific trajectory point pairs), the system improves map accuracy without the need for complex global recalculation.
Solution Approach 2:
The patent changes the parameter being corrected from full six-degree-of-freedom pose estimation to specifically targeting rotation offsets (three rotational parameters). This parameter reduction simplifies the calculation by focusing only on the problematic rotational components that cause overlapping, rather than re-calculating all position and orientation parameters.
Data Source
AI summary
A method is provided. The method includes: obtaining global position and orientation information of a first trajectory point; determining at least one pair of second trajectory points based on the global position and orientation information; obtaining a relative position and orientation estimation information and a relative position and orientation measurement information of each pair of second trajectory points; and calculating a local rotation offset at the first trajectory point based on the relative position and orientation estimation information and the relative position and orientation measurement information of each pair of second trajectory points in the at least one pair of second trajectory points to correct the global position and orientation information.