Synthesizing Probe Data from Overhead Imaging for Map Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The generation and validation of map data for roadways are hindered by the unavailability of probe data due to limitations in access or local laws, which restrict the collection of such data, especially in remote or restricted areas where probe vehicles are sparse.
Innovation Solution
A system that synthesizes probe data from overhead imaging data using a generative neural network, which generates encoded features representing attributes like lane markers and road boundaries, allowing for the creation of lane-level maps and validation of existing data by comparing synthesized and actual probe data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If probe data is collected from actual vehicles traversing roadways, then map data can be generated with high accuracy, but probe data becomes unavailable in regions where local laws prohibit collection or where probe vehicles are sparse
Solution Approach 1:
The system creates synthetic probe data by copying the structure and characteristics of real probe data through a generative neural network. The probe model generates encoded features that represent vehicle trace data and sensor detections, effectively creating artificial copies of probe data that can be used where actual probe data is unavailable due to legal restrictions or sparse vehicle coverage
Solution Approach 2:
The generative neural network acts as an intermediary between overhead imaging data and the mapping pipeline. Instead of directly using unavailable probe data, the system uses the neural network to translate imaging data into synthetic probe data format, bridging the gap between available imaging data and the probe data requirements of existing mapping systems
2Reliability
If multiple passes through a region are used to acquire sufficient trace data for map generation, then map coverage improves, but time consumption and operational complexity increase
Solution Approach 1:
The system performs preliminary synthesis of probe data from overhead imaging data before actual mapping operations. By pre-generating synthetic probe data that represents vehicle traces and sensor detections, the system eliminates the need for multiple actual vehicle passes through the region, as the synthetic data already provides sufficient coverage for map generation
Solution Approach 2:
The mapping system becomes self-sufficient by generating its own probe data from publicly available overhead imaging data. Instead of relying on external probe vehicles to collect data through multiple passes, the system uses the probe model to create all necessary probe data internally, making the mapping process independent of actual vehicle traffic patterns
3Adaptability or versatility
If synthesized probe data is used to generate maps in areas without actual probe data, then mapping capability is extended to restricted areas, but data validation becomes more challenging
Solution Approach 1:
The system implements a feedback mechanism where synthesized probe data is compared against actual probe data when available. This validation process provides feedback to assess the accuracy of the synthetic data, allowing the system to identify discrepancies and improve the reliability of generated maps in both restricted and non-restricted areas
Data Source
AI summary
Systems, methods, and other embodiments described herein relate to improving the generation and validation of map data by synthesizing probe data. In one embodiment, a method includes acquiring imaging data about a roadway, the imaging data being from a remote source. The method includes encoding the imaging data using a probe model to generate features. The method includes generating, from the features using the probe model, probe data that compliments the imaging data for the roadway. The method includes providing the probe data that includes a vehicle trace and detections about attributes of the roadway.