Street-Level Imagery Mining for Intersection Turn Restriction Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic maps do not accurately capture and present turn restrictions at intersections, leading to illegal routes and increased collision risks due to the high cost and inefficiency of manual human review, which does not scale with map expansion or new restrictions.
Innovation Solution
A system that mines map data using street-level imagery and machine learning models to automatically identify and predict turn restrictions by recognizing signs related to turn maneuvers, reducing the need for manual effort and improving map updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual human review of GPS traces and street-level imagery is used to identify turn restrictions, then map accuracy can be maintained, but the cost and time required increases significantly and does not scale with map expansion
Solution Approach 1:
The patent replaces the mechanical system of manual human review with an automated computer vision system using convolutional neural networks (CNNs) to detect and classify turn restriction signs in street-level imagery. The system processes images automatically to identify signs such as 'Do Not Enter' and 'One Way' signs, extracting turn restriction information without human intervention, thereby maintaining accuracy while dramatically improving productivity and scalability.
Solution Approach 2:
The patent uses GPS traces from vehicles as a copy or proxy indicator of turn restriction behavior. By analyzing aggregated GPS data showing where vehicles consistently turn or avoid turning at intersections, the system infers the presence of turn restrictions without directly observing signs, providing an alternative method to validate and supplement visual sign detection.
2Reliability
If manual review processes are used to keep up with new turn restrictions, then map data can be updated, but the scaling capability is insufficient as electronic maps expand to new geographic regions
Solution Approach 1:
The patent implements a self-service system where the automated computer vision pipeline continuously processes new street-level imagery and GPS data to automatically detect and add turn restrictions to the map database. The system serves itself by autonomously identifying signs, classifying restriction types, and updating map data structures without requiring manual review processes, enabling the system to scale reliably as electronic maps expand to new geographic regions.
3Measurement precision
If more intersections are reviewed manually, then turn restriction coverage improves, but the cost and human effort required increases proportionally
Solution Approach 1:
The patent replaces manual human review with automated computer vision processing that can analyze thousands of intersections simultaneously. The CNN-based system processes street-level imagery and GPS traces automatically, detecting turn restriction signs and inferring restrictions from vehicle behavior patterns, thereby achieving comprehensive coverage across all intersections without proportional increases in human effort or time investment.
Data Source
AI summary
Disclosed herein is a system for mining map data to automatically identify and/or predict turn restrictions. The systems uses street-level imagery to recognize and locate posted signs (e.g., physical signs, electronic signs) that signal, or are in some way related to, a turn restriction at an intersection. Then, the techniques use a cascade of machine learning models to accurately predict whether the recognized and located signs impose the turn restriction at the intersection. Consequently, the human effort required to identify turn restrictions is greatly decreased, if not completely eliminated. Furthermore, electronic maps can more efficiently be expanded and/or updated which improves the experience for vehicle drivers that reply upon the electronic maps for directions from original locations to destination locations.