Traffic Speed Estimation Using Sensor and Probe Data Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic speed estimation systems face challenges with accuracy and coverage, as in-road sensors provide comprehensive but noisy data, while probe vehicles offer accurate but limited coverage, leading to incomplete and inaccurate traffic information.
Innovation Solution
Combining data from road sensors and probe vehicles using Bayesian linear regression and machine learning techniques to transform and smooth traffic speed data, enhancing accuracy and coverage by inferring speeds between sensors and applying transforms to improve data quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If road sensors are used for traffic speed estimation, then coverage is comprehensive, but data accuracy deteriorates due to noise and errors
Solution Approach 1:
The patent combines data from road sensors and probe vehicles using a Bayesian framework. Road sensor data provides comprehensive coverage while probe vehicle data provides accurate ground truth measurements. The Bayesian linear regression model merges these two data sources, using the accurate but limited probe data to correct the noisy but comprehensive road sensor data, thereby achieving both wide coverage and high accuracy simultaneously.
2Measurement precision
If probe vehicles are used for traffic speed estimation, then data accuracy is improved, but coverage becomes limited
Solution Approach 1:
The patent uses road sensors as an intermediary to extend the coverage of probe vehicle data. The Bayesian model uses probe vehicle measurements to establish a correction relationship with road sensor readings. This correction model then applies to all road sensor locations, effectively using the accurate probe data as a mediator to improve the accuracy of comprehensive road sensor coverage throughout the network.
3Area of stationary object
If road sensor data is used directly, then coverage is comprehensive, but data quality deteriorates due to noise
Solution Approach 1:
The patent implements a feedback mechanism where probe vehicle measurements provide ground truth feedback to correct road sensor readings. The Bayesian linear regression model continuously learns the relationship between road sensor data and probe vehicle data, using the accurate probe measurements as feedback to adjust and improve the reliability of road sensor data across the entire network.
4Reliability
If probe vehicle data is used directly, then data quality is high, but completeness deteriorates due to limited coverage
Solution Approach 1:
The patent transforms the parameters of road sensor data using a Bayesian linear regression model that is calibrated with probe vehicle data. By changing the parameters (applying correction factors and transformations) to the comprehensive road sensor dataset, the system maintains the completeness of road coverage while improving the quality of all measurements through the learned relationship with accurate probe vehicle observations.
Data Source
AI summary
A computer-implemented method includes obtaining road sensor data reflecting speeds of traffic on road segments, transforming the road sensor data using vehicle probe data for the road segments reflecting vehicle speeds, and producing speed estimates for the road segments using the transformed road sensor data. The method can further include determining speeds for road segments between road sensors by smoothing data from sensors near the road segments.