Driver Drowsiness Notification Using Dual Eye-Closure Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing drowsiness sign notification devices fail to accurately notify drivers of drowsiness due to issues with determining eye closure states, particularly when changes in posture or extraneous light affect the accuracy of face image analysis, leading to interruptions in notification during a scarcely waking state.
Innovation Solution
A drowsiness sign notification device that uses a processor to detect eye-closed states from face images generated at different times, calculates first and second eye closure ratios over specified periods, and notifies the driver when these ratios exceed predetermined thresholds, with adjustments in notification timing based on autonomous driving levels and extraneous light intensity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If PERCLOS is used to detect eye closure state, then drowsiness detection accuracy is improved, but determination reliability deteriorates when posture changes or extraneous light enters the vehicle interior
Solution Approach 1:
The patent dynamically adjusts the eye closure determination threshold based on detection confidence levels. When the determination unit cannot reliably determine eye closure state (low confidence), the system adapts by using alternative metrics or adjusting thresholds, rather than using a fixed threshold approach. This dynamic adaptation maintains reliability under varying conditions like posture changes or extraneous light.
Solution Approach 2:
The system changes the determination parameters (thresholds, time windows, confidence levels) based on environmental conditions and detection quality. When extraneous light or posture changes degrade image quality, the system adjusts the parameters used for eye closure determination, transitioning from strict PERCLOS-based detection to more flexible determination criteria that account for uncertain conditions.
2Reliability
If eye closure determination is interrupted when determination cannot be made, then false positive notifications are reduced, but notification continuity deteriorates during scarcely waking state
Solution Approach 1:
The system performs preliminary assessment of determination confidence before interrupting notification. When the determination unit encounters difficulty in assessing eye closure state, it evaluates whether the uncertainty is temporary or persistent. If temporary, the system maintains notification continuity using previously established drowsiness state, rather than immediately interrupting, thereby preventing false negatives during scarcely waking states.
Solution Approach 2:
The patent implements a cushioning mechanism where the system maintains notification state even when determination confidence is low, rather than immediately switching states. This creates a buffer that prevents premature interruption of valid drowsiness notifications, while still allowing interruption when multiple consecutive determinations fail, thus cushioning against both false positives and false negatives.
3Device complexity
If fixed threshold is used for eye closure ratio, then detection simplicity is improved, but adaptability to different driving conditions deteriorates
Solution Approach 1:
The patent implements dynamic threshold adjustment based on autonomous driving levels and environmental conditions. The determination threshold is not fixed but adapts according to the driving context - for example, using different thresholds for manual driving versus autonomous driving modes, and adjusting for extraneous light conditions. This maintains relative simplicity while achieving adaptability.
Solution Approach 2:
The system changes determination parameters including thresholds, time windows, and confidence levels based on driving conditions. When autonomous driving level increases or extraneous light intensity changes, the system adjusts these parameters accordingly, allowing the same detection system to adapt to various driving scenarios without requiring completely different detection algorithms for each condition.
Data Source
AI summary
A drowsiness sign notification device detects an eye-closed state of a driver of a vehicle from each of face images; determines whether a first eye closure ratio of the number of face images with the eye-closed state to the number of the face images generated in a first period exceeds a first threshold; determines whether a second eye closure ratio of the number of face images with the eye-closed state to the number of the face images generated in a second period longer than the first period exceeds a second threshold less than or equal to the first threshold after determination that the first eye closure ratio does not exceed the first threshold; and notifies detection of a sign of drowsiness to the driver, when the first eye closure ratio exceeds the first threshold or that the second eye closure ratio exceeds the second threshold.