Road Edge Detection Using Traffic Sign Position Cues
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous and semi-autonomous vehicles face challenges in detecting road edges and planes under adverse conditions such as snow, drifting sand, or heavy foliage, leading to unsatisfactory performance or system failure.
Innovation Solution
The implementation of a road edge detection system that recognizes roadside traffic signs using computer vision and stored sign information to estimate the location of the road edge, even when it is not visibly distinct, by focusing on a targeted search area based on sign type and position specifications from jurisdictional data and empirical data aggregated from multiple vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor-based techniques (camera, LIDAR, radar) are used to detect road edge and road plane, then the system can provide automated driving assistance, but the detection performance deteriorates under adverse conditions such as snow, drifting sand, large quantities of leaves, or flooding
Solution Approach 1:
The patent introduces traffic signs as intermediary objects to indirectly determine road edge location. Instead of directly detecting the road edge through sensors (which fails under adverse conditions), the system detects traffic signs that are positioned at known distances from the road edge, using these signs as mediators to infer road edge location even when the road edge itself is obscured by snow, sand, leaves, or flooding
Solution Approach 2:
The system pre-stores sign information including the relationship between different traffic sign types and their corresponding distances to road edges. This preliminary data preparation allows the system to quickly determine road edge location by matching detected sign types against the stored information, without needing to directly observe the road edge in real-time adverse conditions
2Measurement precision
If the system searches the entire image area to detect road edge, then it may find the road edge when visible, but the detection efficiency decreases and processing time increases
Solution Approach 1:
The patent segments the image processing task by first detecting and identifying traffic signs, then using the sign locations and types to define a restricted search area for road edge detection. This segmentation transforms the problem from searching the entire image to searching only the relevant region around detected signs, significantly improving detection efficiency while maintaining accuracy
Solution Approach 2:
The system applies different processing strategies to different regions of the image: traffic sign detection is performed across the entire image, while road edge detection is focused only in local regions around the detected signs. This local quality approach concentrates computational resources where they are most needed, improving both efficiency and precision
Data Source
AI summary
In some examples, one or more processors of a vehicle may store, in a computer-readable medium, sign information relating a sign type of a traffic sign to a distance to a road edge for a plurality of sign types. The one or more processors may receive at least one image from at least one sensor, and may recognize a sign type of a traffic sign in the at least one image. Further, the one or more processors may determine a likely location of a road edge in relation to the vehicle based at least partially on the recognized sign type and the stored sign information. In some examples, the one or more processors may send one or more control signals to at least one component of the vehicle based on the determined likely location of the road edge relative to the vehicle.