Sliding Window Map-Matcher for Real-Time GPS Probe Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current map-matching algorithms for GPS probe measurements face challenges such as high computational cost, outlier filtering, and accuracy issues, particularly in real-time traffic processing and on stacked roadways, limiting their deployment in global-scale real-time systems.
Innovation Solution
The Sliding Window Map-Matcher (SWMM) approach reduces computational complexity by retaining probability estimates and filtering outliers, using past and future states to infer road positions, and limiting candidate locations, enabling efficient and accurate matching of probe measurements to paths in geographic locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional map-matching algorithms are used to match GPS probe measurements to road segments, then matching accuracy can be maintained, but computational cost becomes excessively high for real-time processing
Solution Approach 1:
The algorithm segments the road network into discrete road segments with pre-calculated geometric properties and candidate location sets. By dividing the continuous road network into manageable segments and pre-computing candidate locations for each segment, the algorithm reduces the computational burden during real-time probe data processing while maintaining matching accuracy.
Solution Approach 2:
The algorithm performs preliminary actions by pre-calculating and storing candidate location sets for each road segment before probe data arrives. This includes pre-computing geometric properties, potential matching locations, and filtering criteria. When probe data is received, the system only needs to evaluate pre-computed candidates rather than performing full map-matching calculations in real-time.
2Measurement precision
If comprehensive candidate locations are considered for matching, then matching accuracy improves, but computational complexity increases
Solution Approach 1:
The algorithm applies local quality by creating specialized candidate location sets tailored to each road segment's characteristics. Each segment has its own set of candidate locations based on local geometric properties, curvature, and spatial relationships. This localized approach ensures high accuracy for each segment while avoiding the need to evaluate all possible locations across the entire road network.
Solution Approach 2:
The algorithm extracts and retains only the most relevant candidate locations for each road segment, discarding improbable matches. By filtering out unlikely candidate locations based on geometric constraints and spatial reasoning, the system maintains high matching accuracy while significantly reducing the number of candidates that require computational evaluation.
3Productivity
If probe measurements are processed individually, then processing speed increases, but accuracy decreases due to inability to filter outliers and handle stacked roadways
Solution Approach 1:
The algorithm merges information from multiple probe measurements by maintaining and updating candidate location sets across sequential measurements. By combining evidence from multiple measurements and using temporal consistency checks, the system can filter outliers and resolve ambiguities on stacked roadways while still processing measurements in a streamlined fashion.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An apparatus for matching probe measurements to a path in a geographic location includes a receiver, a window manager, a location generator, a path calculator, and an output. The receiver is configured to receive a stream of probe measurements. The window manager is configured to fill a window with the measurements, to select an additional measurement from the stream, and to select an oldest measurement in the window. The location generator is configured to generate candidate locations for the measurements in the window and the additional measurement. The path calculator is configured to match the oldest measurement to a candidate location. The output is configured to output a path-matched probe measurement based on the oldest measurement and the candidate location matched to the oldest measurement.