Intersection Trajectory Generation Using Road Marking Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating travel trajectory data in intersections for autonomous driving are inaccurate due to large GPS variations and deviate from actual vehicle trajectories, especially in complex intersection shapes, limiting their practicality.
Innovation Solution
A vehicle-mounted device equipped with a road marking recognizer and a travel trajectory data generator that uses captured image data to recognize road markings, such as line markings, specific colors, striped areas, and diamond-shaped markings, to accurately generate travel trajectory data for autonomous driving systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If GPS-based trajectory estimation is used in intersections, then trajectory data can be generated without special-purpose vehicles, but the accuracy deteriorates due to large GPS variations
Solution Approach 1:
The patent introduces road markings as an intermediary element to bridge the gap between GPS-based positioning and accurate trajectory determination. By detecting and tracking road markings (lane lines, stop lines, crosswalks) in intersection areas, the system creates a reference framework that corrects GPS variations and provides precise trajectory information without requiring special-purpose vehicles or expensive sensors.
Solution Approach 2:
The patent replaces the mechanical/GPS-based trajectory estimation system with an optical detection system that uses cameras to capture and analyze road markings. This substitution eliminates reliance on GPS satellite signals and their associated inaccuracies in urban canyons and intersections, providing continuous and accurate position information through visual recognition of painted road features.
2Device complexity
If conventional trajectory generation methods are used, then data collection is simplified, but the trajectory data deviates from actual vehicle paths in complex intersections
Solution Approach 1:
The patent performs preliminary detection and recognition of road markings in advance of trajectory generation. By pre-identifying lane lines, stop lines, and other intersection markings before the vehicle reaches critical decision points, the system establishes an accurate geometric model of the intersection layout, enabling reliable trajectory prediction and path planning even in complex intersection configurations.
3Measurement precision
If special-purpose vehicles with expensive sensors are used, then measurement precision improves, but cost and human work requirements increase
Solution Approach 1:
The patent replaces expensive, complex sensor systems (laser scanners, multiple cameras, specialized equipment) with a simple, inexpensive camera that captures standard images of road markings. This approach uses readily available, low-cost imaging technology to achieve accurate trajectory data collection, eliminating the need for specialized vehicles and expensive sensor suites while maintaining measurement precision.
Data Source
AI summary
In a device for generating travel trajectory data in an intersection, a road marking recognizer is configured to recognize a road marking in the intersection using captured image data of the intersection. A travel trajectory data generator is configured to, in response to a result of recognition by the road marking recognizer, generate travel trajectory data that is data enabling specification of a travel trajectory in the intersection for autonomous driving.