Vehicle Position Estimation Using Lane Mark Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle position estimation systems face challenges in accurately determining the lateral position of a vehicle within a lane due to limitations in map data and satellite reception conditions, and require dedicated modules for recognizing multiple painted lines, which limits their applicability and accuracy.
Innovation Solution
A vehicle position estimation apparatus that includes an image information acquiring unit, a painted line recognizing unit, a painted line recognition result processing unit, a map information acquiring unit, a vehicle lateral position calculating unit, and an on-map vehicle position calculating unit, which together estimate the vehicle's position by analyzing camera images and map data to detect lane changes and calculate lateral positions, enabling accurate lane-level positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dead reckoning with GNSS coordinate information, vehicle speed signal information, and angular velocity information is used to calculate vehicle position, then the vehicle position can be estimated, but the lateral position accuracy deteriorates with satellite reception conditions such as multipath causing errors of about 10 m
Solution Approach 1:
The patent introduces painted line recognition as an intermediary element to bridge the gap between satellite positioning and actual vehicle position. By detecting painted lines (lane markings) in the environment and using their known positions from map data, the system creates a reference framework that mediates the inaccurate satellite positions to achieve accurate lateral positioning within lanes.
2Area of stationary object
If a navigation map representing roads with links, nodes, and attribute information is used for map matching, then road-level position can be determined, but lane-level detailed shape information is lost making it impossible to estimate the travel lane accurately
Solution Approach 1:
The patent applies local quality by enhancing the map data with lane-level detailed shape information specifically at the vehicle's current location. Instead of requiring complete high-detail maps everywhere, the system enriches map data locally around the vehicle with painted line positions and lane boundaries, allowing accurate lane estimation only where needed while maintaining overall map coverage.
3Measurement precision
If a dedicated painted line recognition module is used to detect multiple painted lines for accurate lane positioning, then lateral position accuracy improves to tens of centimeters, but the device complexity and applicability deteriorate due to limited compatibility with existing systems
Solution Approach 1:
The patent makes the painted line recognition capability universal by integrating it into the existing satellite positioning system rather than requiring a separate dedicated module. The same image processing unit that processes satellite positioning data also processes painted line detection, allowing one system to perform multiple functions (satellite positioning and painted line recognition) without adding significant complexity.
4Ease of operation
If only two painted lines immediately adjacent to the vehicle are detected for lane-keeping application, then the system works for basic lane keeping, but the applicability deteriorates when more than two painted lines are needed for vehicle position estimation
Solution Approach 1:
The patent implements a dynamic painted line recognition system that automatically adapts to different scenarios. The system dynamically determines how many painted lines to detect and process based on the application requirements - detecting only two lines for basic lane-keeping or multiple lines for comprehensive vehicle position estimation. This dynamic adaptability allows the same system to serve multiple purposes without being overwhelmed by complexity.
Data Source
AI summary
A vehicle position estimation apparatus according to the present invention includes a painted line recognition result processing unit configured to calculate whether the vehicle has made a lane change and a direction of the lane change when the vehicle has made the lane change, a vehicle lateral position calculating unit configured to calculate a lane where the vehicle is in travel and a vehicle lateral position that is a lateral position of the vehicle in the lane based on whether the vehicle has made the lane change, the direction of the lane change, and a lane connection relation, and an on-map vehicle position calculating unit configured to calculate an on-map position of the vehicle on the road based on a map information, the lane where the vehicle is in travel, the vehicle lateral position, and a forward travel distance.