Digital Road Model Generation via Trajectory-Image Overlay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating digital road models are inefficient, leading to inaccurate and outdated models due to the limited number of vehicles collecting data, inaccuracy of positioning sensors, and difficulties in merging aerial or satellite images with varying resolutions and distortions, which affects the precision of navigation and autonomous vehicle systems.
Innovation Solution
A system and method that combine vehicle trajectories with aerial or satellite images to create high-precision digital road models by overlaying trajectories onto images, analyzing driving-relevant features within a defined corridor, and generating accurate models using statistical methods and image processing techniques to align and merge images effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a small number of specially equipped vehicles are used to collect and update road data, then the complexity of the data collection system is reduced, but the productivity of updating road data deteriorates significantly
Solution Approach 1:
The invention enables ordinary vehicles to automatically contribute their own trajectory and sensor data to the digital road model without requiring specialized equipment or manual intervention. Each vehicle serves itself by generating and transmitting data that benefits the collective system, eliminating the need for a dedicated fleet of data collection vehicles.
Solution Approach 2:
The system allows any standard vehicle equipped with basic sensors (cameras, radar, lidar, GPS) to participate in data collection. The same vehicle infrastructure serves both its primary transportation function and the secondary function of contributing to digital road model updates, maximizing resource utilization without requiring specialized vehicles.
2Device complexity
If positioning sensors with limited accuracy are used in data collection vehicles, then the cost and complexity of the system are reduced, but the manufacturing precision of the digital road model deteriorates
Solution Approach 1:
The invention merges data from multiple independent vehicles traversing the same road segment. By combining trajectories and sensor observations from many vehicles, the system achieves high-precision digital road models through statistical aggregation, compensating for the limited accuracy of individual positioning sensors.
Solution Approach 2:
The system uses the collected trajectory data and sensor observations as feedback to continuously refine and update the digital road model. The model is iteratively improved by comparing actual vehicle paths and sensor detections against the model predictions, correcting accumulated errors and maintaining high accuracy over time.
3Area of stationary object
If aerial or satellite images with varying resolutions and distortions are stitched together, then the coverage area of the digital road model is expanded, but the measurement precision of the model deteriorates due to discontinuities at seams
Solution Approach 1:
The invention uses vehicle trajectories as an intermediary reference framework to align and integrate aerial or satellite images. Instead of directly stitching images with varying resolutions and distortions, the system anchors image features to the more accurate vehicle-derived trajectory data, which serves as a mediating coordinate system that resolves discrepancies between different image sources.
4Reliability
If complex image stitching and road recognition algorithms are used to process aerial images, then the completeness of road feature detection is improved, but the computing time and processing complexity increase significantly
Solution Approach 1:
The system performs preliminary actions by collecting and storing vehicle trajectory and sensor data in real-time as vehicles traverse road segments. This pre-collected data serves as a foundation that constrains and guides subsequent image processing, reducing the search space and computational requirements for road feature detection in aerial images.
Solution Approach 2:
The invention segments the road mapping problem into independent road segment units, each processed using local vehicle data and corresponding aerial image portions. This segmentation allows parallel processing of multiple segments and reduces the overall computational burden compared to processing entire large-scale aerial images globally.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for creating a digital road model for at least one road segment comprising the receiving of at least one trajectory of a vehicle for the at least one road segment in a vehicle-external database and to the receiving of at least one image showing at least parts of the at least one road segment, wherein the image has a perspective that corresponds to an image captured substantially vertically downward form an elevated position. The at least one trajectory is superposed with the at least one image in such a way that the at least one trajectory corresponds with the course of a road in the at least one image. The at least one image is analyzed in a corridor that extends along the trajectory and that contains the trajectory in order to identify driving- or positioning-relevant features of the road segment in the corridor. From the driving- or positioning-relevant features that were identified in the at least one image oriented on the basis of the at least one trajectory and in the corridor containing the trajectory, a digital road model is generated.