Driver Gaze Uncertainty Mapping for Distraction Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional driver monitoring systems using gaze estimation face challenges in situations like strong sunlight, dark sunglasses, and high head angles, which degrade gaze determination accuracy, resulting in binary quality signals that are not granular enough for reliable driver distraction assessment.
Innovation Solution
A system that processes images of a driver to determine a gaze region with a probability distribution, including a gaze point and uncertainty value, and associates it with pre-defined attentive and inattentive regions around the vehicle, providing a driver distraction level and system degradation level, enabling more granular assessment and outputting signals for distraction and system reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional binary quality signal classification is used for gaze determination, then the system is simple to operate, but the measurement precision of driver distraction assessment is insufficient
Solution Approach 1:
The patent transforms the binary quality signal (usable/not usable) into a continuous probability distribution with mean (gaze point) and standard deviation (gaze uncertainty). This parameter transformation enables granular assessment of driver distraction by providing nuanced confidence levels rather than simple binary classifications, directly improving measurement precision while maintaining computational efficiency through standard statistical parameters.
2Adaptability or versatility
If gaze estimation is performed in challenging conditions (strong sunlight, dark sunglasses, high head angles), then the system maintains operational capability, but the reliability of gaze determination degrades
Solution Approach 1:
The patent implements feedback through the gaze uncertainty value, which quantifies the reliability of each gaze estimation. When conditions are challenging (strong sunlight, dark sunglasses, high head angles), the system naturally produces higher uncertainty values, providing feedback about the quality of the measurement. This allows the system to maintain operational capability while honestly reporting reduced reliability, enabling downstream systems to adjust their trust in the gaze data accordingly.
3Ease of operation
If binary quality signal classification is used, then the ease of operation is high, but the granularity of driver distraction assessment is insufficient
Solution Approach 1:
The patent adds a new dimension to the gaze quality assessment by introducing the probability distribution framework. Instead of a single binary dimension (usable/not usable), the system now operates in a two-dimensional space defined by mean (gaze point) and standard deviation (gaze uncertainty). This dimensional expansion preserves ease of operation through simple statistical parameters while dramatically increasing information granularity, allowing differentiation between various levels of confidence in gaze estimation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Method and apparatus, including computer programs, for providing a driver distraction signal and/or a system degradation signal indicating a reliability of the driver distraction signal, based on a gaze of a driver of a vehicle. Images of the driver of the vehicle are received, which contain information indicating a gaze of the driver. For each image, a gaze region for the driver is determined, including a gaze point and a gaze uncertainty value. For each image, the gaze region is associated with a region among several regions around the vehicle, wherein the regions include at least one pre-defined attentive region and at least one pre-defined inattentive region. Based on the determined gaze region, the gaze uncertainty value, and the region for a plurality of images, a driver distraction level and/or system degradation level indicating a reliability of the driver distraction level are determined. A driver distraction signal and/or a system degradation signal are output, indicative of the determined driver distraction level and/or the system degradation level.