Autonomous Vehicle Localization Using LIDAR Overlap Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately determining their position due to differences in sensor perspectives and heights between the vehicle and the mapping vehicle, leading to reduced overlap scores and potential failure in recognizing their location, which can cause warnings or reliance on less accurate positioning methods.
Innovation Solution
The system generates an overlap score by calculating the match between the vehicle's sensor data and map data, accounting for differences in sensor positions and occlusions, and uses ray tracing to include non-matching points if they intersect with the map data, thereby improving the accuracy of position determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vehicle's sensor data is directly compared with map data, then the position determination can be performed quickly, but the accuracy is reduced due to sensor perspective differences and occlusions
Solution Approach 1:
The patent transforms the 2D sensor data into a 3D point cloud representation, adding a vertical dimension to account for sensor height differences. This dimensional transformation enables accurate matching between vehicle sensor data and map data by considering the third dimension (height) in the spatial coordination system, thereby resolving the accuracy issue caused by perspective differences.
Solution Approach 2:
The patent introduces an intermediary processing step that generates a point cloud from sensor data and performs ray tracing to identify occluded regions. This intermediary representation acts as a mediator between the raw sensor data and the map data, enabling accurate comparison by accounting for occlusions and perspective differences through the point cloud transformation.
2Reliability
If sensor height differences are not accounted for, then the processing is simpler, but the overlap score decreases leading to position recognition failure
Solution Approach 1:
The patent applies local quality transformation by adjusting the coordinate system specifically for the vertical dimension where sensor height differences exist. Instead of uniformly transforming all coordinates, the system selectively applies height compensation to the relevant vertical components of the sensor data, maintaining simplicity in non-affected dimensions while improving reliability in the height-critical positioning aspect.
Solution Approach 2:
The patent changes the vertical parameter (height) in the coordinate transformation process to account for sensor mounting height differences. By modifying the Z-coordinate parameter in the spatial transformation, the system compensates for height variations between vehicle sensors and map acquisition sensors, thereby improving position recognition reliability without requiring complete system redesign.
3Measurement precision
If occluded regions are not considered, then the overlap calculation is faster, but the position determination accuracy is compromised
Solution Approach 1:
The patent performs preliminary identification of occluded regions through ray tracing before conducting the final overlap score calculation. By pre-processing the sensor data to mark occluded areas in advance, the system avoids time-consuming exclusion checks during the main matching process, thereby maintaining high overlap score accuracy while minimizing additional processing time.
Solution Approach 2:
The patent extracts and separately processes occluded regions from the main sensor data stream. By identifying and isolating occluded points through ray tracing, the system removes these problematic data points from the overlap calculation, ensuring that only valid, non-occluded points contribute to the position determination, thus maintaining accuracy without requiring reprocessing of the entire dataset.
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
This approach enhances the vehicle's ability to accurately determine its position and navigate by improving the overlap score representation, reducing errors caused by sensor perspective differences and occlusions, and ensuring reliable autonomous driving.
Implementation Method 1
position information from three-dimensional light detection and ranging (LIDAR) data captured by a LIDAR sensor
Implementation Method 2
position information from three-dimensional light detection and ranging (LIDAR) data captured by a LIDAR sensor
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Disclosed are systems, methods, and computer-readable storage media to control a vehicle. In one aspect, a method includes capturing point-cloud data representative of a surrounding of an autonomous vehicle with one or more LIDAR sensors, identifying a point in the point cloud data as a non-matching point in response to the point having no corresponding point in a map used to determine a position of the autonomous vehicle, determining whether the non- matching point is to be used in a determination of an overlap score based on one or more comparisons of the point cloud data and the map, determining the overlap score in response to the determining whether the non-matching point is to be used in the determination of the overlap score, determining a position of the autonomous vehicle based on the overlap score and the map, and controlling the autonomous vehicle based on the position.