LiDAR Road Surface Modeling for Non-Planar Vehicle Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road-modeling techniques for automated vehicles and automotive safety systems are inadequate as they often rely on a single plane to represent travel surfaces, which fail to accurately predict vehicle behavior on non-planar surfaces with varying angles and slopes.
Innovation Solution
A road-model definition system that uses point-cloud data from a LiDAR unit to tessellate the travel surface into cells, fitting 2D planes to each cell to create a 3D piece-wise linear model, allowing for optimal vehicle operation by dividing or combining cells based on deviation and parallel thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single plane is used to represent the travel surface, then the road model is simple, but the accuracy of vehicle behavior prediction on non-planar surfaces deteriorates
Solution Approach 1:
The patent divides the travel surface into multiple discrete cells through tessellation, with each cell represented by a separate plane. This segmentation allows the road model to capture local variations in surface geometry (crowns, slopes, curves) while maintaining computational efficiency. Each cell's plane is determined independently based on ground points within that cell, enabling accurate representation of non-planar surfaces without requiring a single complex surface model.
2Measurement precision
If complex non-linear techniques are used to characterize the travel lane shape, then the road model accuracy improves, but the data processing complexity increases
Solution Approach 1:
The patent transforms the road modeling approach by changing the mathematical representation from complex non-linear continuous surfaces to simple linear planes discretized across multiple cells. This parameter change allows using basic linear equations instead of complex non-linear functions, significantly reducing computational complexity while maintaining sufficient accuracy for vehicle control applications. The piece-wise linear model captures essential road geometry features without requiring sophisticated mathematical models.
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 reduces data processing complexity and improves vehicle control by providing a more accurate representation of the travel surface, enhancing the predictability of vehicle behavior on uneven road surfaces.
Implementation Method 1
receives point-cloud data from a Light Detection And Ranging unit (lidar-unit) indicative of an area proximate to a host-vehicle
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A road-model-definition system (10) suitable for an automated-vehicle includes a lidar-unit (14) and a controller (20). The lidar-unit (14) is suitable to mount on a host-vehicle (12). The lidar-unit (14) is used to provide a point-cloud (16) descriptive of an area (18) proximate to the host-vehicle (12). The controller (20) is in communication with the lidar-unit (14). The controller (20) is configured to: select ground-points (26) from the point-cloud (16) indicative of a travel-surface (28), tessellate (36) a portion of the area (18) that corresponds to the travel-surface (28) to define a plurality of cells (38), determine an orientation (42) of each cell (38) based on the ground-points (26) within each cell (38), define a road-model (40) of the travel-surface (28) based on the orientation (42) of the cells (38), and operate the host-vehicle (12) in accordance with the road-model (40).