RNN Human Trust Modeling for Vehicle Take-Over Intent Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Human trust calibration in interactions with automated vehicle systems is challenging due to its abstract and context-dependent nature, leading to potential over-trust or under-trust, which can result in misuse or disengagement from automation.
Innovation Solution
A computer-implemented method using a recurrent neural network (RNN) based human trust model that analyzes inputs from vehicle operations and driving scenes, incorporating crowd-sourced data to predict short-term and long-term trust states and determine a driver's take-over intent, thereby adapting automation behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human supervision and intervention are required in automated vehicle systems, then safety and reliability are improved, but system complexity and operational burden increase
Solution Approach 1:
The system implements a feedback mechanism where the trust model continuously monitors driver trust levels and provides feedback to the automation system. This allows the system to adapt its behavior based on real-time trust assessments, maintaining safety while reducing unnecessary complexity through intelligent adaptation rather than constant human supervision
Solution Approach 2:
The automation system performs self-assessment of driver trust levels through the trust model, enabling it to autonomously adjust its operation mode without requiring continuous human evaluation. The system serves itself by automatically calibrating to driver trust states, reducing operational burden while maintaining reliability
2Reliability
If the automation system adapts its behavior to driver trust levels, then trust calibration is improved, but computational complexity and data processing requirements increase
Solution Approach 1:
The trust model acts as an intermediary layer between the driver and the automation system. Instead of the automation system directly processing complex driver state information, the trust model mediates by interpreting driver behavior and translating it into trust level assessments, simplifying the computational burden on the main automation system while achieving accurate trust calibration
Solution Approach 2:
The system replaces direct mechanical observation and interpretation of driver trust with a computational trust model that uses machine learning algorithms. This substitution allows for more efficient processing of trust-related data compared to traditional methods, improving trust calibration while managing computational complexity through algorithmic optimization
3Measurement precision
If the system collects and analyzes crowd-sourced survey data, then prediction accuracy is improved, but data processing time and computational resources increase
Solution Approach 1:
The system performs preliminary processing of crowd-sourced survey data during off-peak periods or in parallel computation streams. By pre-processing and organizing this data before it is needed for real-time trust assessment, the system reduces the computational burden during critical driving moments, maintaining prediction accuracy while minimizing data processing time during operation
Solution Approach 2:
The data processing task is segmented into multiple independent processing streams that can be executed in parallel. The trust model divides the analysis of crowd-sourced data into discrete computational units, allowing simultaneous processing of different data subsets, thereby reducing overall processing time while maintaining comprehensive analysis accuracy
4Speed
If the trust model predicts take-over intent in real-time, then responsiveness is improved, but computational load and processing speed requirements increase
Solution Approach 1:
The trust model implements local quality assessment by focusing computational resources on predicting specific critical parameters like take-over intent rather than analyzing all possible driver states uniformly. By identifying and prioritizing the most relevant trust indicators for real-time prediction, the system achieves high responsiveness while reducing overall computational load through selective analysis
Data Source
AI summary
A system and method for providing an RNN-based human trust model that include receiving a plurality of inputs related to an autonomous operation of a vehicle and a driving scene of the vehicle and analyzing the plurality of inputs to determine automation variables and scene variables. The system and method also include outputting a short-term trust recurrent neural network state that captures an effect of the driver's experience with respect to an instantaneous vehicle maneuver and a long-term trust recurrent neural network state that captures the effect of the driver's experience with respect to the autonomous operation of the vehicle during a traffic scenario. The system and method further include predicting a take-over intent of the driver to take over control of the vehicle from an automated operation of the vehicle during the traffic scenario.