User State Inference Control Using Bayesian Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human-machine interaction methods fail to accurately identify user emotional or physiological states with sufficient reliability and determine appropriate device actions, leading to suboptimal user experience and safety.
Innovation Solution
A method using sensors to obtain data on users and their environments, applying Bayesian inference and causal models to determine user states and their causes, iteratively refining these determinations until a condition is met, and then controlling devices based on these states and causes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If multiple indicators such as facial expressions, speech tone, and biological data are used for emotion recognition, then the detection capability is improved, but the reliability of user state identification deteriorates due to inability to accurately determine appropriate actions
Solution Approach 1:
The patent segments the complex user state identification process into distinct iterations, where each iteration focuses on identifying a specific state and its cause. This segmentation allows the system to break down the complex task of monitoring multiple indicators into manageable steps, improving both detection capability and reliability by systematically processing each state independently.
Solution Approach 2:
The patent implements a dynamic iterative process where the system continuously monitors user state indicators and adapts its analysis based on changing conditions. The iterative nature allows the system to dynamically adjust its detection and identification processes, refining its accuracy over time while maintaining responsiveness to changing user states.
2Measurement precision
If iterative processing is used to determine user states and their causes, then the accuracy of state identification is improved, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-defining the structure of iterations and the relationships between states and causes before actual processing begins. The system establishes the framework for iterative analysis in advance, which allows for more efficient processing during execution while maintaining high accuracy through systematic refinement.
Solution Approach 2:
The iterative process incorporates feedback mechanisms where each iteration's results inform subsequent iterations. The system uses the outcomes of previous analyses to refine its approach, allowing it to converge on accurate state identification more efficiently and reduce overall processing time through learned optimizations.
Data Source
Figure 1A
Figure 1B
Figure 2A~3
AI summary
A computer-implemented method for controlling a device, said method comprising: • obtaining (S10), using sensors (SR), data (DT) representative of said user (Us) and of an environment (Env); • determining (S20), based on said data, indicators (Em_I) of states of said user and at least one event indicator (Ev_I); • determining (S30), based on said state indicators and said data, using a Bayesian inference model (BIM), probabilities (Em_P) of states of said user (Us); and • determining (S40), based on said probabilities and said at least one event indicator, at least one state (Em_S) of said user and at least one event (Ev_C) causing the user to be in said at least one state; • determining (S50), based on said at least one state and event, a control action (Ac) of said device.