Multi-Layer Map Positioning for Overpass LiDAR Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional positioning solutions for autonomous driving, such as LiDAR-based systems, struggle with accuracy in complex scenarios like overpasses and multi-layer roads due to insufficient height information representation, leading to positioning errors and drifts.
Innovation Solution
A multi-layer Gaussian model map is used to process point cloud data, incorporating height information, to determine the estimated position, height, and posture of objects with high accuracy by employing a histogram filter for probability estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR-based positioning solution is used, then positioning capability is provided, but positioning accuracy deteriorates in complex scenarios like overpasses and multi-layer roads due to insufficient height information representation
Solution Approach 1:
The patent introduces multiple height layers (first height layer, second height layer, etc.) to represent different elevation levels in the map data. This multi-dimensional approach allows the system to distinguish between objects at different heights, such as vehicles on ground level versus vehicles on overpasses, thereby resolving positioning ambiguities in complex multi-layer road scenarios
Solution Approach 2:
The patent applies different processing strategies to different height layers. For example, when point cloud data matches the first height layer, the system determines the object is on the ground; when it matches the second height layer, the system determines the object is on an overpass. This localized processing for different spatial regions improves positioning accuracy in specific complex scenarios
2Reliability
If conventional single-layer map representation is used, then device complexity is reduced, but positioning reliability deteriorates due to inability to distinguish multi-level road structures
Solution Approach 1:
The patent segments the map representation into multiple height layers, where each layer contains point cloud data and positioning information for a specific elevation level. This segmentation allows the system to process and analyze positioning data from different heights independently, improving reliability in multi-level scenarios while keeping each individual layer's complexity manageable
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The disclosure provides a method for positioning an object, comprising: obtaining a map (1021) related to a region (1011) where the object is located, the map including a plurality of map layers having different height information; determining, based on the map and current point cloud data (1022) related to the object, an estimated position of the object, an estimated height corresponding to the estimated position and an estimated probability that the object is located at the estimated position with an estimated posture; and determining, at least based on the estimated position, the estimated height and the estimated probability, positioning information for the object, the positioning information indicating at least one of a current position of the object, a current height of the object and a current posture of the object. Satellite view (1010) shows a complex overpass scene. Visualization effect image (1020) has a multi-layer map (1021) for region (1011) where the object is located, and a visualization effect image of the current point cloud data (1022). The map (1021) and the current point cloud data (1022) are inputted into a histogram filter (1030), and then processed to obtain a matched result. Visualization effect image (1040) is a combination of (i) a reflection information matching histogram (1031) for a set of map layers (1033), and (ii) a height information matching histogram (1032) for the set of map layers (1034). A visualized positioning result (1050) is obtained which reflects the positioning information of the object.