Driver Trust Prediction for Real-Time Autonomous Vehicle Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately modeling driver trust in autonomous systems and implementing corrective actions without frustrating the driver during data collection.
Innovation Solution
A system and method using a Selective Windowing Attention Network (SWAN) to predict driver trust by processing multi-modal time-series data from vehicle and occupant sensors, and automatically modifying vehicle systems based on the predicted trust levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If driver trust is accurately modeled from measurable data, then prediction accuracy is improved, but system complexity increases
Solution Approach 1:
The trust prediction system is segmented into multiple independent components: data collection module, feature extraction module, neural network prediction module, and action determination module. Each component processes specific aspects of trust modeling independently, reducing overall system complexity while maintaining prediction accuracy through modular architecture.
Solution Approach 2:
The patent introduces intermediary elements including trained neural network models that act as mediators between raw sensor data and trust predictions, and lookup tables storing pre-determined actions that mediate between trust values and system responses. These intermediaries simplify the complex mapping relationships without sacrificing prediction accuracy.
2Reliability
If corrective actions are implemented based on trust predictions, then driver trust is improved, but risk of frustrating the driver increases
Solution Approach 1:
The system performs preliminary actions by determining and storing multiple candidate corrective actions in advance for different trust prediction scenarios. When a low trust prediction occurs, the system selects from pre-prepared actions rather than generating responses in real-time, reducing the risk of inappropriate responses that could frustrate the driver while maintaining trust-building effectiveness.
Solution Approach 2:
The patent implements feedback mechanisms where the system monitors driver responses to corrective actions and uses this information to adjust future trust predictions and action selections. This closed-loop feedback ensures that corrective actions are adapted to individual driver preferences and contexts, minimizing frustration while maximizing trust improvement.
3Reliability
If multiple vehicle systems are modified in real-time, then driver trust is enhanced, but control complexity increases
Solution Approach 1:
The patent creates a universal trust prediction and control system that can modify multiple different vehicle systems (cruise control, steering, braking, entertainment, climate control) through a single integrated architecture. The determination of corrective actions and modification of vehicle systems are performed by a unified control mechanism that adapts to different system types, reducing overall control complexity despite the multi-system scope.
Data Source
AI summary
A method and system for modeling trust levels of drivers and modifying autonomous systems in real time in response to predicted trust levels. These modifications may be made by one or more systems, including an autonomous driving agent. An end-to-end attention network known as a Selective Windowing Attention Network (SWAN) learns directly from time-series data and assigns attention to critical areas.