Generative Model-Driven Sampling for Low-Power User-State Sensing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

XR devices face limitations in utility and availability of sensor information due to limited battery life and significant power requirements for always-on assistance, necessitating a framework for efficient sensor sampling based on user state uncertainty.

Innovation Solution

A generative model-driven framework that selects sensor sampling modes based on user state uncertainty, switching between high-power and low-power modes to balance accuracy and power consumption, using biosensors and environmental sensors for adaptive sensing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If sensor sampling is performed at high frame rates using high-power sensors, then measurement precision and user state prediction accuracy are improved, but power consumption increases

Engineering Contradiction:
Improveuser state prediction accuracyVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system dynamically adjusts sensor sampling modes based on predicted user state uncertainty. When uncertainty is high, the system switches to high-power, high-precision sampling modes. When uncertainty is low, it transitions to low-power modes. This dynamic adaptation resolves the contradiction by making power consumption variable rather than constant, matching resource usage to actual prediction needs.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes sampling parameters (frame rate, sensor activation) based on the uncertainty level of predicted user states. The generative model outputs uncertainty metrics that directly control sampling parameter selection, allowing the system to optimize between precision and power consumption by adjusting parameters according to actual prediction confidence levels.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If sensor sampling is performed continuously at high power modes, then reliability of user state measurement is improved, but duration of battery operation deteriorates

Engineering Contradiction:
Improveuser state measurement reliabilityVSAvoidbattery life
Core Design Contradiction:
ReliabilityVSDuration of action of stationary object

Solution Approach 1:

The system implements periodic assessment of user state uncertainty using the generative model, and adjusts sampling modes based on these periodic evaluations. Rather than continuous high-power sampling, the system periodically checks prediction uncertainty and only activates high-power sensors when necessary, extending battery life while maintaining measurement reliability when needed.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The generative model provides feedback about prediction uncertainty that controls sensor sampling decisions. This feedback loop ensures that high-power sampling is activated only when the model indicates high uncertainty about user state, maintaining measurement reliability while avoiding unnecessary power consumption during high-confidence prediction periods.

Inventive Principle:
Principle #23Feedback

3Use of energy by moving object

If low-power sensor sampling modes are used, then power consumption is reduced, but measurement precision and prediction accuracy deteriorate

Engineering Contradiction:
Improvepower consumptionVSAvoiduser state prediction accuracy
Core Design Contradiction:
Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The system dynamically switches between low-power and high-power sampling modes based on real-time uncertainty assessment. The generative model continuously evaluates prediction uncertainty, and the sampling system adapts its power consumption level accordingly, using low-power modes during high-confidence predictions and switching to high-power modes when uncertainty increases.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes sampling parameters (frame rate, sensor activation) based on uncertainty level. When the generative model indicates low uncertainty, the system maintains low-power sampling parameters. When uncertainty exceeds thresholds, parameters are adjusted to higher precision modes, optimizing the balance between power consumption and prediction accuracy.

Inventive Principle:
Principle #35Parameter changes

4Adaptability or versatility

If high-power sensors are activated for always-on assistance, then adaptability of user assistance is improved, but power consumption increases

Engineering Contradiction:
Improvecontextual awarenessVSAvoidpower consumption
Core Design Contradiction:
Adaptability or versatilityVSUse of energy by moving object

Solution Approach 1:

The system performs periodic uncertainty assessment using the generative model and adjusts sensor activation accordingly. Instead of continuous high-power operation, the system periodically evaluates whether contextual awareness is needed based on prediction uncertainty, activating high-power sensors only during periods when adaptability requirements justify the power consumption.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The generative model provides feedback about user state prediction confidence that controls sensor activation decisions. This feedback mechanism enables the system to maintain adaptability by activating sensors when prediction uncertainty indicates a need for updated information, while conserving power during periods when current predictions remain accurate.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4641356A1Generative model-driven sampling for adaptive sparse multimodal sensing of user environment and intent
Publication Date: 2025.10.29 META PLATFORMS TECHNOLOGIES LLC
  • EP4641356A1 patent drawingFigure 1
  • EP4641356A1 patent drawingFigure 2
  • EP4641356A1 patent drawingFigure 3

AI summary

The disclosed computer-implemented method may include (1) predicting a user state, wherein the user state is measurable via a plurality of different sensor sampling modes, (2) determining a level of uncertainty associated with the predicted user state, and (3) selecting, from the plurality of different sensor sampling modes, a sampling mode to measure the user state. Selecting the sampling mode may include selecting a first sampling mode in response to determining that the level of uncertainty is above a threshold or selecting a second sampling mode in response to determining that the level of uncertainty is below the threshold. Various other methods, systems, and computer-readable media are also disclosed.