3D Road Surface Segmentation for Elevated Area Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for three-dimensional road area segmentation, such as those using Digital Elevation Maps and deep convolutional encoder-decoder architectures, are less effective for detecting elevated surfaces like roundabouts and traffic isles, and often rely too heavily on context, leading to inaccuracies.
Innovation Solution
A method for three-dimensional road area segmentation that uses input data with height values from depth maps to classify elements into road and elevated surface classes, employing DBSCAN or a density-based multiple surface aggregation algorithm, and dynamically adjusts thresholds based on standard deviation to improve accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional methods use Digital Elevation Maps and deep convolutional encoder-decoder architectures for road area segmentation, then the processing framework is established, but the detection accuracy for elevated surfaces like roundabouts and traffic isles deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the road area classification into distinct categories: road surfaces, elevated surfaces (sidewalks, traffic isles, roundabouts), and obstacles. This is achieved through height-based classification where elements are categorized into road class, elevated surface class, or obstacle class based on their height values relative to the road surface, improving detection accuracy for elevated surfaces
Solution Approach 2:
The patent transitions from two-dimensional image processing to three-dimensional segmentation by incorporating height values from depth maps. This dimensional extension allows the system to distinguish elevated surfaces from road surfaces based on their vertical position, resolving the limitation of conventional 2D methods in detecting roundabouts and traffic isles
2Measurement precision
If conventional methods rely heavily on context for detection, then the classification framework is established, but the measurement precision deteriorates due to over-reliance on contextual information
Solution Approach 1:
The patent extracts and utilizes height information from depth maps as a primary feature for classification, separating this critical measurement from contextual dependencies. By focusing on the extracted height values rather than relying heavily on contextual cues from surrounding elements, the system achieves more precise classification of road surfaces, elevated surfaces, and obstacles
Solution Approach 2:
The patent changes the classification parameter from context-based semantic understanding to height-based physical measurement. By using height values (e.g., elements above a certain threshold are classified as elevated surfaces or obstacles) as the primary classification criterion, the system achieves more precise and reliable detection independent of contextual variations
3Reliability
If density-based clustering algorithms are used for surface aggregation, then the segmentation is performed, but the sensitivity to outliers increases
Solution Approach 1:
The patent applies local quality by performing density-based clustering locally for each class (road class, elevated surface class, obstacle class) separately rather than globally. This localized approach allows the algorithm to adapt to the specific density characteristics of each surface type, improving segmentation reliability while reducing sensitivity to outliers that may have different density patterns in different regions
Data Source
AI summary
A method for a three-dimensional road area segmentation for a vehicle. The method includes: obtaining input data including multiple elements representing a scene of an environment of the vehicle, the input data indicating heights of the scene and resulting at least partially from a sensor detection of the environment; carrying out the three-dimensional road area segmentation based on the input data. The three-dimensional road area segmentation includes: classifying the elements based on the heights into at least a road class and an elevated surface class, thereby providing a classification result for each of the elements; determining a density level for multiple of the elements based on the classification results, the density levels being based on the density of the different classes; identifying at least a road surface and an elevated surface in the scene by forming the surfaces based on the density levels.