Road Shape Recognition via Radar and Camera Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional road shape recognition systems face accuracy issues when detecting roadside objects discontinuously, such as with trees or buildings, and struggle when white lines are faded or absent, leading to decreased recognition accuracy.
Innovation Solution
A method combining on-board radar and camera systems to recognize road edges and lane shapes, where the recognition results from both sensors are compared and compensated for each side of the vehicle to improve accuracy, using reliability levels to determine the coincidence and identify the road shape.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If road shape recognition uses only radar detection of roadside objects, then the system can operate without white lines, but recognition accuracy significantly decreases when roadside objects are discontinuous (trees, buildings) or absent (embankments)
Solution Approach 1:
The patent combines radar detection results and camera detection results to recognize road shapes. The radar detects roadside objects and determines road edge shapes, while the camera detects white lines and determines lane shapes. These two independent detection systems are merged through a determination unit that compares their results and an identification unit that synthesizes them into a final road shape identification, thereby maintaining high accuracy across various road conditions.
2Measurement precision
If road shape recognition uses only camera detection of white lines, then the system achieves high accuracy when white lines are clear, but recognition fails when white lines are faded or absent
Solution Approach 1:
The patent merges radar-based road edge shape detection with camera-based lane shape detection. When white lines are clear, the camera provides accurate lane shape data. When white lines are faded or absent, the radar-detected road edge shape compensates for the camera's inability to detect lane shapes, ensuring continuous and accurate road shape recognition under all conditions.
3Measurement precision
If the system uses both radar and camera detection results, then recognition accuracy improves through compensation, but the device complexity increases
Solution Approach 1:
The patent segments the road shape recognition task into two independent modules: radar-based road edge shape determination and camera-based lane shape determination. Each module operates independently with its own processing logic, and their results are combined through a determination unit that compares coincidence and an identification unit that synthesizes the final result. This segmentation allows the system to leverage the strengths of each sensor while managing complexity through modular architecture.
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 recognition accuracy by compensating for difficult detection situations between radar and camera systems and improves the identification of road shapes, even when detection is challenging on one side, thereby reducing errors and maintaining high accuracy.
Implementation Method 1
a transmission wave, such as an optical wave or a millimeter-wave is radiated over a predetermined angle ahead of or to the side of the vehicle and a reflection wave thereof is received
Implementation Method 2
a transmission wave, such as an optical wave or a millimeter-wave is radiated over a predetermined angle ahead of or to the side of the vehicle and a reflection wave thereof is received
Implementation Method 3
an image captured by an on-board camera
Data Source
AI summary
A method and an apparatus are provided to recognize a shape of a road on which a vehicle is traveling. Road edge shapes on a left side and a right side of the vehicle are recognized, from positions of roadside objects detected based on detection information from an on-board radar. Lane shapes that are shapes of lane boundary lines on the left side and the right side of the vehicle are recognized, from positions of lane boundary lines detected based on an image captured by an on-board camera. For each of the left side and the right side, such that the recognized road edge shape and the recognized lane shape are compared with each other, and the road shape is identified based on the comparison results.