Road Shape Recognition Using Stereo Camera and Stationary Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road shape recognition systems face challenges in accurately detecting continuous road shapes, selecting effective stationary objects, and detecting curves before entry, especially in environments with numerous three-dimensional objects or without visible road features like white lines or curbs, leading to delayed control or warning timing and precision issues.
Innovation Solution
A road shape recognition device that captures images ahead using stereo cameras to detect road regions and estimate road shapes by applying road shape models, correcting estimates based on three-dimensional boundary points, and utilizing a combination of image processing and sensor data to improve precision and responsiveness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If stationary objects are selected based on previous road shape processing results, then the road shape estimation process is simplified, but the precision deteriorates when the road shape is discontinuous or previously erroneously estimated
Solution Approach 1:
The patent applies preliminary action by detecting stationary objects before road shape estimation and creating multiple candidate sets based on different hypotheses (with and without previous results). This allows the system to prepare multiple preparation paths in advance, then select the most appropriate one based on current road conditions, thereby maintaining precision while managing complexity.
Solution Approach 2:
The system changes the parameter of object selection criteria dynamically - sometimes using previous road shape results as selection criteria, other times using alternative criteria when discontinuities or errors are detected. This parameter change allows the system to adapt to different road conditions and maintain estimation precision across varying scenarios.
2Measurement precision
If multiple road shapes are hypothesized and stationary objects are selected for each hypothesis, then the selection process becomes more thorough, but the processing time and complexity increase significantly
Solution Approach 1:
The patent applies partial action by not fully processing all hypothetical road shapes with complete object selection. Instead, it creates candidate sets with different levels of processing thoroughness and selects the appropriate level based on current conditions, reducing processing time while maintaining sufficient accuracy for the given scenario.
Solution Approach 2:
The system performs preliminary classification of road shape hypotheses and prepares corresponding object candidate sets in advance. By organizing hypotheses and their associated objects beforehand, the system avoids redundant processing during real-time operation, significantly reducing processing time while maintaining selection accuracy.
3Device complexity
If only stationary objects effective for road shape estimation are selected using yaw rate and steering angle sensors, then the system works with limited data, but curves cannot be detected before entry when no lateral acceleration occurs
Solution Approach 1:
The patent applies preliminary action by detecting stationary objects and potential curve features before the vehicle actually enters the curve. By using image processing to identify road geometry changes in advance, the system can prepare curve detection information prior to lateral acceleration occurring, enabling earlier and more reliable curve detection.
Solution Approach 2:
The system introduces stationary objects detected through image processing as an intermediary between the vehicle's motion sensors and the road shape estimation. These stationary objects serve as reference points that provide additional information about upcoming curves, complementing the limited data from yaw rate and steering angle sensors.
4Ease of manufacture
If road shape detection relies on visible road features like white lines and curbs, then the detection process is straightforward, but detection fails on roads without such features
Solution Approach 1:
The patent applies universality by designing a detection system that performs multiple functions: it can detect road shapes using visible features like white lines and curbs when available, and simultaneously detect stationary objects and road geometry when such features are absent. This multi-functional approach enables the system to adapt to various road environments while maintaining a relatively simple implementation.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A road shape recognition device capable of accurately recognizing the road shape of a road that lies ahead in the travel direction of a vehicle is provided. A road shape recognition device 1 detects a road region of the road based on an image capturing a scene ahead in the travel direction of the vehicle, and estimates the shape of the road based on that road region. Thus, it is possible to accurately recognize road shapes over distances ranging from short to long.