LiDAR Static Map Construction via Dynamic Point Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic objects in the environment interfere with the accuracy of spatial map construction for self-driving vehicles, leading to inaccurate localization and navigation.
Innovation Solution
A method using LiDAR sensors to generate 3D frames, identifying dynamic points by distance threshold, and correcting them as static points through comparison with previous frames, ensuring accurate static map construction for self-driving vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If LiDAR sensor scans the three-dimensional space continuously to build the static map, then the map coverage and completeness are improved, but dynamic objects are included in the map causing inaccuracy
Solution Approach 1:
The patent extracts and removes dynamic points from the LiDAR frame by comparing them with the static map. Points that show significant distance deviations are identified as dynamic objects and removed, leaving only static points for map updates. This extraction process resolves the contradiction by eliminating harmful dynamic elements while preserving the beneficial static map data.
Solution Approach 2:
The patent implements a feedback mechanism where the static map is continuously updated with corrected static points from new LiDAR frames. The system uses the existing static map as a reference to identify and remove dynamic points, then feeds the cleaned static points back into the map. This closed-loop feedback ensures the map remains accurate despite continuous scanning through dynamic environments.
2Productivity
If distance threshold method is used to identify dynamic points, then the processing speed is improved, but the precision of dynamic object detection may be reduced
Solution Approach 1:
The patent uses distance threshold as a parameter to identify dynamic points by comparing the distance between LiDAR points and their corresponding static map points. By adjusting the threshold parameter, the system can balance between processing speed and detection precision. This parameter-based approach resolves the contradiction by providing a controllable method that adapts to different operational requirements.
Solution Approach 2:
The patent performs preliminary comparison of each LiDAR point with the static map before final classification. By pre-calculating distances and identifying potential dynamic points early in the processing pipeline, the system prepares data for subsequent refinement steps. This preliminary action enables faster processing while maintaining detection accuracy through subsequent verification.
3Measurement precision
If all points in LiDAR frame are compared with static map points, then the detection accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the LiDAR frame into individual points and processes them separately by finding corresponding points in the static map. This segmentation allows the system to handle complex comparisons in a modular fashion, reducing overall computational complexity while maintaining detection accuracy for each point.
Solution Approach 2:
The static map serves as an intermediary reference structure that enables efficient comparison of LiDAR points. Instead of directly comparing all LiDAR points with each other or with raw sensor data, the system uses the pre-built static map as a mediator to identify dynamic points. This intermediary approach simplifies the computational process while maintaining detection precision.
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
The method effectively removes dynamic objects from the static map, providing a cleaner and more accurate map for localization and navigation, enhancing the reliability of self-driving vehicle operations.
Implementation Method 1
A three-dimensional space is scanned by a LiDAR sensor to generate a LiDAR frame including multiple points in the three-dimensional space in a time sequence
Data Source
Figure 1~2
Figure 3~4
Figure 5A~5B
AI summary
A method and a system for spatial static map construction are provided. In the method, a three-dimensional space is scanned by using a LiDAR sensor to generate a LiDAR frame including multiple points in the three-dimensional space in a time sequence. As for each point in the LiDAR frame, a corresponding point closest to the point is found from a static map built according to the three-dimensional space, and a distance from the corresponding point is calculated. The point is labelled as a dynamic point if the distance is greater than a threshold, and otherwise labelled as a static point. Each labelled dynamic point is compared with points in N LiDAR frames generated before the time sequence, and corrected as a static point if included in the N LiDAR frames. The dynamic points in the LiDAR frame are removed, and each static point is updated to the static map.