Smartphone Motion Classifier Using Confidence Thresholds

Solution Overview

Problem

Existing motion classification methods for mobile devices are inefficient in accurately transitioning between different motion states, particularly due to high false positives and toggling issues, which can lead to inaccurate logs in applications like geo-fencing and health monitoring.

Innovation Solution

A method utilizing triaxial accelerometer data to determine motion states with statistical techniques like Gaussian mixture models, where transitions are delayed based on confidence levels to reduce toggling and allow simultaneous active states, such as Biking and Stationary, to account for temporary changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If motion state transitions are triggered immediately when confidence level exceeds threshold, then responsiveness to motion changes is improved, but false positives and toggling issues increase

Engineering Contradiction:

Improveresponsiveness to motion changesVSAvoidaccuracy of motion state classification

Core Design Contradiction:

SpeedVSReliability

Solution Approach 1:

The system performs preliminary verification by checking if the confidence level exceeds a threshold before triggering a state transition. This preliminary check prevents immediate transitions on spurious signals, reducing false positives while maintaining responsiveness to genuine motion changes.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The confidence level acts as an intermediary mechanism between raw sensor data and state transitions. By introducing this intermediate verification layer, the system filters out noisy transitions while preserving legitimate state changes, thereby reducing toggling issues without sacrificing responsiveness.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If motion state transitions are delayed to reduce toggling, then false positives are reduced, but responsiveness to legitimate motion changes decreases

Engineering Contradiction:

Improveaccuracy of motion state classificationVSAvoidresponsiveness to motion changes

Core Design Contradiction:

ReliabilityVSSpeed

Solution Approach 1:

The system performs preliminary verification by checking if the confidence level exceeds a threshold before triggering a state transition. This preliminary check prevents immediate transitions on spurious signals, reducing false positives while maintaining responsiveness to genuine motion changes.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The confidence level acts as an intermediary mechanism between raw sensor data and state transitions. By introducing this intermediate verification layer, the system filters out noisy transitions while preserving legitimate state changes, thereby reducing toggling issues without sacrificing responsiveness.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Device complexity

If only single motion states are recognized at a time, then state classification simplicity is maintained, but accuracy in complex motion scenarios decreases

Engineering Contradiction:

Improvestate classification simplicityVSAvoidaccuracy of motion state classification

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The system changes the parameter of state recognition from single-state to multi-state by allowing multiple motion states to be active simultaneously. This is achieved by maintaining separate confidence levels for different states and enabling transitions when their respective thresholds are met, thereby improving accuracy in complex scenarios like biking while stationary.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system dynamically adapts the number of active states based on the motion scenario. Rather than fixing the system to a single state, it allows the state configuration to change dynamically according to sensor inputs, enabling accurate representation of complex motions where multiple states coexist.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS11029328B2Smartphone motion classifier

Publication Date: 2021.06.08 QUALCOMM INC

AI summary

Disclosed is an apparatus and method for classifying a motion state of a mobile device comprising: determining a first motion state associated with a highest probability value and with a first confidence level greater than a first threshold; entering the first motion state; while the first motion state is active, determining a second motion state associated with a highest probability value and with a second confidence level greater than the first threshold, the second motion state being different from the first motion state; determining whether the second motion state is to be entered; and in response to determining that the second motion state is to be entered, entering the second motion state.