Non-Uniform Increment Sampling for Vehicle Position Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solid-state laser radar systems have a short sensing distance and issues with occlusion, leading to sparse or missing point cloud clusters for networked vehicles, making accurate position estimation challenging, especially for vehicles far away or partially occluded.
Innovation Solution
An independent non-uniform increment sampling method is used to generate virtual mapping points based on spatiotemporal aligned image and point cloud data, filling sparse regions by sampling according to point density, and reversing these points to the original point cloud space for enhanced point cloud completion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If solid-state laser radar is used to sense networked vehicle position, then sensing cost is reduced, but sensing distance becomes short and point cloud clusters become sparse or missing due to occlusion
Solution Approach 1:
The patent introduces an intermediary filling algorithm that uses image data as a mediator to generate virtual point clouds. These virtual point clouds serve as filler data to compensate for the sparse or missing regions in the original laser radar point clouds, thereby improving point cloud completeness without changing the radar hardware
Solution Approach 2:
The patent creates virtual point cloud copies by mapping image features into three-dimensional space. These copied point cloud data represent the occluded or distant vehicle regions and are merged with the original point clouds to restore completeness
2Adaptability or versatility
If neural network-based perception algorithm is used for position estimation, then generalization ability is improved, but accuracy deteriorates for targets with serious occlusion or sparse point clouds
Solution Approach 1:
The patent performs preliminary filling of sparse point cloud regions before the neural network processes the data. By pre-generating virtual point clouds to complete the occluded regions, the input data to the neural network is improved, enabling accurate position estimation for previously problematic cases
3Ease of operation
If uniform sampling is used for point cloud processing, then processing simplicity is maintained, but sampling efficiency deteriorates in sparse regions
Solution Approach 1:
The patent applies different sampling strategies to different regions of the point cloud based on their local characteristics. Dense regions use one sampling approach while sparse regions use another, optimizing sampling efficiency for each local area rather than applying a uniform approach throughout
Data Source
AI summary
The present application discloses a method and a device for estimating the position of a networked vehicle based on independent non-uniform increment sampling. By mapping a laser radar point cloud to a spatiotemporal aligned image, independent non-uniform increment sampling is carried out on the mapping points falling in an advanced semantic constraint region of the image according to a point density of the depth interval where the mapping points are located, and the virtual mapping points generated by sampling are reversely mapped to the original point cloud space and merged with the original point cloud, and the combined point cloud is used to estimate the position of the networked vehicle based on a deep learning method, so as to solve the inaccurate position estimation problem of sheltered or remote networked vehicles due to the sparseness or missing of its own point cloud clusters.