Road Shape Recognition Using Moving Vehicle Trajectories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road shape recognition methods for vehicles fail to accurately calculate the road shape, especially in complex scenarios like interchanges, ramps, and areas with few roadside objects, due to incorrect detection of stationary objects and curvature radius estimation.
Innovation Solution
A method that recognizes the road shape by determining the difference in curvature between the actual road and the estimated road shape, adjusting lane probability calculations based on this difference, and using a transmission wave to acquire object-unit data for effective road edge recognition, even in situations with limited roadside objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the curve radius is corrected based on recognized road shape, then the lane probability calculation is improved, but the correction becomes inaccurate when the vehicle does not travel along the recognized road shape (e.g., at interchanges, ramps, or bus stops)
Solution Approach 1:
The patent changes the parameter used for road shape recognition from stationary objects to moving objects (other vehicles). By detecting the positions and movements of surrounding vehicles, the system calculates the road shape dynamically, which remains accurate even when the vehicle is at interchanges, ramps, or bus stops where stationary objects may be misleading or absent.
2Loss of information
If stationary objects are used to recognize road edges, then the road shape can be estimated, but the detection fails when roadside objects are occluded by preceding vehicles or when the absolute number of roadside objects is small
Solution Approach 1:
The patent uses other moving vehicles as intermediary objects to infer road shape information. Instead of directly detecting road edges through stationary objects, the system detects the positions and trajectories of surrounding vehicles, which naturally follow the road geometry, thereby indirectly obtaining accurate road shape data even when direct road edge detection is compromised.
3Ease of operation
If the vehicle state (steering angle, yaw rate) is used to calculate curve radius, then the calculation is straightforward, but the state of turn does not conform to the actual road shape
Solution Approach 1:
The patent introduces feedback by using the detected positions and movements of other vehicles to continuously refine and correct the road shape estimation. The system compares the inferred road geometry from multiple vehicle trajectories against the vehicle's actual turn state, adjusting the curve radius calculation to better conform to the real road shape rather than relying solely on steering angle and yaw rate.
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 the accuracy of road shape recognition, preventing errors in curvature correction and improving the detection of road edges, even in challenging conditions, thereby ensuring more correct and frequent recognition of the road shape.
Implementation Method 1
a distance/angle measuring device 5 for measuring a distance to an object ahead of the vehicle 1 and an angle in a width direction of the vehicle 1
Data Source
AI summary
An example of recognition of the shape of a road where a vehicle travels is provided. An object type as to whether an object is a moving or stationary object is determined according to a relative speed of the object and a speed of the vehicle. Object-unit data effective for recognizing a road shape is extracted according to the determination. The object-unit data is used for forming data of a roadside object group, based on which a road edge is recognized. A series of the processes is repeatedly performed at a predetermined cycle. After the extraction process, a data addition process is performed to add object-unit data obtained in the extraction process of the previous cycle to object-unit data obtained in the extraction process of the present cycle. In a recognition process, a road edge is recognized according to the object-unit data obtained in the data addition process.