Lane Localization via Multi-Camera Fusion and GPS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current GPS navigation systems lack the accuracy to precisely locate a vehicle within a specific lane of a road, especially during disruptions in satellite signals, which limits their ability to provide timely lane changes and driver assistance.
Innovation Solution
A system and method that capture and fuse multiple overlapping images of a 360-degree field of view to detect lane markings, combine this data with GPS signals and vehicle state information to generate a global lane map, and localize the vehicle within its lane, while also projecting paths of travel to activate driver assistance warnings for potential deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional GPS navigation systems are used to locate the vehicle, then the system can provide general location information, but the accuracy is insufficient to locate the vehicle within a specific lane
Solution Approach 1:
The system divides the road environment into multiple lanes by detecting and identifying lane markings in the captured image. The lane detection module segments the road surface into distinct lanes based on visual features, allowing the system to determine which specific lane the vehicle is in, thereby achieving lane-level location accuracy without requiring a single complex positioning device
Solution Approach 2:
The patent introduces an intermediary lane detection and identification system that bridges the gap between GPS coordinates and actual vehicle position. By using lane markings as intermediate reference features, the system translates general GPS location into precise lane-level positioning, enabling accurate lane localization while maintaining system feasibility
2Reliability
If the system waits for GPS satellite signals to determine vehicle location, then the positioning can be obtained, but during signal disruptions the vehicle cannot be accurately located
Solution Approach 1:
The system performs preliminary lane detection and mapping by capturing images and identifying lane markings in advance, before GPS signal disruptions occur. The lane detection module pre-processes visual information to establish the vehicle's current lane position, so that when GPS signals are interrupted, the system can continue to provide accurate lane-level positioning using the pre-acquired lane structure information
Solution Approach 2:
The system continuously monitors both GPS satellite signals and visual lane markings, using feedback from multiple sources to maintain positioning reliability. When GPS signals are disrupted, the feedback loop switches to relying on lane marking detection and inertial sensor data, ensuring continuous and reliable vehicle location information without system failure
3Measurement precision
If the system provides lane-level localization accuracy, then earlier lane change directions can be provided, but the system complexity increases
Solution Approach 1:
The captured image serves multiple functions: it is used for lane marking detection, lane identification, vehicle positioning, and navigation guidance. By making the image capture system multi-functional, the patent avoids adding separate dedicated sensors for each function, thereby achieving lane-level localization accuracy while minimizing the increase in system complexity through versatile use of existing components
Data Source
AI summary
A method and system for simultaneously generating a global lane map and localizing a vehicle in the generated global lane map is provided. The system includes a plurality of image sensors adapted to operatively capture a 360-degree field of view image from the host vehicle for detecting a plurality of lane markings. The image sensors include a front long-range camera, a front mid-range camera, a right side mid-range camera, a left side mid-range camera, and a rear mid-range camera. A controller communicatively is coupled to the plurality of image sensors and includes a data base containing reference lane markings and a processor. The processor is configured to identify the plurality of lane markings by comparing the detected lane markings from the 360-degree field of view image to the reference lane markings from the data base and to fuse the identified lane markings into the global lane map.