Road Object Position Correction Using Camera and Vehicle Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle systems that rely on High Definition (HD) maps for navigation face inefficiencies and inaccuracies due to limited raw data availability, fewer specialty vehicles for data collection, and delayed updates, leading to errors in road object positioning.
Innovation Solution
A vehicle system that utilizes sensor fusion to correct real-time camera-based estimated positions of road objects by combining image sensor data from cameras with vehicle sensor data from Global Navigation Satellite System (GNSS), LIDAR sensors, and other input devices, using error models and regression analysis to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If third party suppliers use predetermined schedules for updating HD maps, then map updates are performed systematically, but real-time corrections of road object positions cannot be achieved
Solution Approach 1:
The system transitions from static predetermined update schedules to dynamic real-time updates by continuously receiving image sensor data from multiple vehicles and processing deviations from HD map positions immediately when detected, allowing the map to adapt dynamically to current road conditions
Solution Approach 2:
The system performs preliminary detection and correction of road object position deviations by multiple vehicles before official map updates are needed, maintaining current and accurate map data through continuous preliminary corrections rather than waiting for scheduled updates
2Ease of manufacture
If camera-based estimation is used for road object positioning, then the system can operate without specialized equipment, but positioning accuracy is insufficient
Solution Approach 1:
The system merges image sensor data from multiple vehicles with HD map data and uses sensor fusion techniques to combine camera-based estimates with vehicle sensor data, achieving high positioning accuracy using standard camera equipment rather than specialized LIDAR or GPS systems
Solution Approach 2:
The system implements feedback by continuously comparing camera-based road object position estimates with HD map positions, calculating deviations, and using these deviations to correct future estimates through error models, progressively improving accuracy through iterative refinement
3Measurement precision
If error models are updated by removing outlier intersections and determining mean, then the accuracy of road object position correction is improved, but computational complexity increases
Solution Approach 1:
The system extracts and removes outlier intersections from the error model by identifying and eliminating extreme values that would skew the mean calculation, thereby improving the accuracy of position corrections while managing computational complexity through selective data processing
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
The system achieves accurate and real-time correction of road object positions, enhancing navigation accuracy and coverage, especially in areas with obstructed aerial views, and allowing for more frequent and precise map updates.
Implementation Method 1
The input devices include a LIDAR sensor for collecting ranging sensor data associated with an offset distance between the vehicle and the road object
Implementation Method 2
The vehicle system includes a Global Navigation Satellite System (GNSS)
Data Source
AI summary
A vehicle system is provided for correcting in real-time a camera-based estimated position of a road object. The system includes a camera for generating an image input signal including image sensor data associated with the road object. The system further includes one or more input devices for generating a vehicle input signal including vehicle sensor data associated with a position, a speed, and a heading of the vehicle. The system further includes a computer, which includes one or more processors and a non-transitory computer readable medium (CRM) storing instructions. The processor is programmed to match the image sensor data and the vehicle sensor data to one another based on a common time of collection. The processor is further programmed to determine an error model and a deviation of a current camera-based position from a predicted position. The processor is further programmed to update the error model based on the deviation.