HD Map Construction via Landmark Pose Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pose optimization methods for constructing high-definition (HD) maps in autonomous driving do not consider the semantic interpretation of point cloud data, leading to inaccurate and inefficient map construction and updating.
Innovation Solution
A system and method that jointly optimizes pose information of multiple local HD maps and landmarks by identifying and matching data frames associated with landmarks, using sensors like LiDAR and GPS/IMU, to establish robust and accurate HD map construction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If current pose optimization methods are used for constructing HD maps, then the map construction process is simplified, but the accuracy and precision of the HD map deteriorates
Solution Approach 1:
The patent combines pose optimization with semantic interpretation of point cloud data by integrating landmark detection and matching processes. The system jointly optimizes pose parameters and landmark associations, merging previously separate processing steps into a unified framework that improves accuracy while maintaining computational efficiency.
Solution Approach 2:
The patent introduces landmarks as intermediary elements that mediate between raw point cloud data and pose estimation. By detecting and matching landmarks across multiple frames, the system creates reliable correspondence points that improve pose optimization accuracy without significantly increasing computational complexity.
2Device complexity
If semantic interpretation of point cloud data is not considered in pose optimization, then the computational complexity is reduced, but the optimization accuracy deteriorates
Solution Approach 1:
The patent segments the point cloud data processing by identifying and extracting landmark features as distinct semantic elements. This segmentation allows the system to focus computational resources on key feature points rather than processing all points uniformly, improving pose optimization accuracy while controlling computational complexity through selective processing.
3Ease of manufacture
If GPS odometry is used for pose estimation, then the system is simple to implement, but the positioning precision deteriorates under poor GPS signal conditions
Solution Approach 1:
The patent uses landmarks as intermediary reference points that provide reliable positioning information independent of GPS signals. By matching landmarks across multiple frames and using them as correspondence points in pose optimization, the system maintains high positioning precision even when GPS signals are poor or unavailable.
Solution Approach 2:
The patent implements a feedback mechanism where detected landmarks provide corrective information to the pose estimation process. The landmark matching results feed back into the pose optimization, allowing the system to correct GPS odometry errors and maintain accurate positioning without requiring complex alternative systems.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Improves the accuracy and robustness of HD map construction, enabling centimeter-level precision even with decimeter-level GPS positioning, by establishing many-to-many constraints between landmarks and point cloud data frames.
Implementation Method 1
sensor data acquired of a target region by at least one sensor equipped on a vehicle
Implementation Method 2
The odometry of the vehicle may be acquired by estimating the pose of the vehicle using a Global Positioning System (GPS) receiver and one or more Inertial Measurement Unit (IMU) sensors
Data Source
AI summary
Embodiments of the disclosure provide systems and methods for updating an HD map. The system may include a communication interface configured to receive sensor data acquired of a target region by at least one sensor equipped on a vehicle as the vehicle travels along a trajectory via a network. The system may further include a storage configured to store the HD map. The system may also include at least one processor. The at least one processor may be configured to identify a plurality of data frames associated with a landmark, each data frame corresponding to one of a plurality of local HD map on the trajectory. The at least one processor may be further configured to jointly optimize pose information of the plurality of local HD maps and pose information of the landmark. The at least one processor may be further configured to construct the HD map based on the based on the pose information of the plurality of local HD maps.