Traffic Congestion Extension Prediction From Existing Camera Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional congestion prediction services struggle to predict the extension length of sudden congestion accurately due to limited training data and high costs associated with camera installations, and they often provide information after congestion has occurred.
Innovation Solution
An information processing apparatus that classifies and integrates congested sections using geohash and azimuth analysis, sets extension scales based on day-of-the-week, working attribute, and time zone divisions, and predicts extension length using selected determination divisions to account for deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cameras are installed in various places to acquire sudden congestion states, then the ability to detect congestion improves, but the cost increases
Solution Approach 1:
The patent uses congestion images captured by existing cameras as training data to create a digital model that predicts sudden congestion. Instead of installing additional cameras, the system copies and processes image data from existing infrastructure, replacing physical expansion with informational replication.
Solution Approach 2:
The patent replaces the mechanical system of installing physical cameras with an information processing system that uses machine learning models. The mechanical approach (adding hardware) is substituted with an informational approach (processing existing data through algorithms).
2Measurement precision
If congestion states are acquired from congestion images, then the accuracy of congestion state acquisition improves, but the time delay increases
Solution Approach 1:
The patent performs preliminary actions by using congestion images as training data to build prediction models before actual congestion occurs. The system learns patterns from historical image data and applies them to predict future congestion, rather than waiting to detect congestion after it happens.
Solution Approach 2:
The patent introduces an intermediary element: the machine learning prediction model. This intermediary processes the relationship between historical congestion images and future congestion states, enabling indirect prediction rather than direct observation of current congestion conditions.
3Reliability
If conventional congestion prediction methods are used based on steady traffic volume increase, then the prediction reliability for regular congestion improves, but the adaptability to sudden congestion deteriorates
Solution Approach 1:
The patent changes the fundamental parameter used for prediction from steady traffic volume increase to sudden changes in congestion patterns. By training models on congestion images that capture abrupt changes rather than gradual trends, the system adapts to predict sudden congestion while maintaining reliability through rigorous training on relevant data.
Solution Approach 2:
The patent transitions from static prediction based on historical averages to dynamic prediction that responds to changing congestion patterns. The machine learning models are trained to recognize and adapt to varying congestion behaviors, including sudden changes, making the system dynamic rather than static.
Data Source
AI summary
An information processing apparatus classifies each congested section represented by training data including a start point position of congestion and an end point position of the congestion for each congestion direction, recursively repeats processing of connecting adjacent congested sections as a continuous integrated congested section for each congested section to determines which congested sections constitute the integrated congested section, classifies the integrated congested section according to determination divisions for each integrated congested section representing the same congestion, and sets an extension scale of the integrated congested section for each integrated congested section representing the same congestion and for each determination division.