Driver State Estimation Using Gaze Search Behavior Indicators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques struggle to accurately distinguish between a driver's inattentive state and other abnormal states such as disease or aging, leading to erroneous estimations.
Innovation Solution
A driver state estimation apparatus that utilizes a controller to analyze travel environment information and the driver's line of sight, calculating an inattentive probability using a sigmoid function based on multiple indicators of search behavior, and correcting feature values to account for environmental factors like gradient, curvature, illuminance, and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques focus on single indicator of driver's line of sight movement, then the detection method is simple, but the estimation accuracy deteriorates due to inability to distinguish inattentive state from persistent abnormal states
Solution Approach 1:
The patent segments the driver state estimation into multiple independent indicators (line of sight movement frequency, movement duration, saccade characteristics, fixation patterns) rather than relying on a single indicator. Each indicator is evaluated separately and then integrated through the sigmoid function to compute the overall inattentive probability, enabling more accurate distinction between temporary inattentive states and persistent abnormal states.
Solution Approach 2:
The patent transitions from single-dimensional assessment to multi-dimensional assessment by incorporating multiple indicators of search behavior. The sigmoid function integrates these multi-dimensional features (frequency, duration, amplitude, patterns) into a unified probability score, adding analytical depth and enabling better discrimination between different driver states.
2Reliability
If conventional techniques use single indicator for driver state estimation, then the processing time is short, but the reliability deteriorates due to erroneous estimations
Solution Approach 1:
The patent performs preliminary processing of each indicator by extracting specific features (frequency, duration, amplitude) before integration. The sigmoid function is pre-configured with appropriate weights for each indicator, allowing rapid computation. This preliminary structuring of data and computation reduces the time penalty associated with multi-indicator analysis while improving reliability.
Solution Approach 2:
The patent transforms raw line of sight data into multiple derived parameters (frequency, duration, amplitude, patterns) and then integrates them through the sigmoid function. This parameter transformation enables more reliable estimation by capturing different aspects of driver behavior, while the mathematical integration maintains computational efficiency.
3Measurement precision
If conventional techniques rely on single indicator of line of sight movement, then the system complexity is low, but the measurement precision deteriorates due to abnormal behavior patterns
Solution Approach 1:
The patent creates a universal estimation framework using the sigmoid function that can process multiple different indicators (line of sight movement, saccade characteristics, fixation patterns) through a single unified mathematical model. This multi-functional approach improves measurement precision by considering various aspects of driver behavior while maintaining system coherence through the universal sigmoid function.
Solution Approach 2:
The patent implements feedback mechanisms where the computed inattentive probability is continuously monitored and compared against thresholds. When the probability exceeds the threshold for a predetermined time period, the system triggers state estimation. This feedback loop refines the measurement precision by confirming sustained abnormal patterns rather than transient fluctuations, while the feedback structure itself adds computational efficiency.
Data Source
AI summary
A driver state estimation apparatus includes a control circuit configured to estimate whether the driver is in a first state based on travel environment information and the driver's line of sight. The control circuit determines a feature value xi (i=1, . . . , n) for each of plural indicators of search behavior changed according to the driver's state based on the travel environment information and the driver's line of sight. The acquired feature values xi and a preset weight coefficient ai for each of the feature values xi, are used to calculate a first probability p representing a probability that the driver is in a first state, and estimates that the driver is in the inattentive state when a state where the calculated inattentive probability p is equal to or higher than a predetermined value continues for a predetermined time or longer.