Matrix Pressure Sensor Neural Network Calibration

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing matrix pressure sensors face challenges in accurately measuring pressure distribution due to inhomogeneous pixel responses and coupling effects between neighboring pixels, requiring complex calibration processes that are time-consuming and not easily generalizable.

Innovation Solution

A calibration process for matrix pressure sensors involving a network of neurons that processes sensor responses, utilizing a database of real homogeneous support data and simulated partial support data to correct for pixel coupling and non-linearity, allowing for simple, fast, and generalizable calibration across various types of matrices.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a classic calibration method with multiple known force levels is applied to each pixel, then measurement precision is improved, but calibration time increases significantly

Engineering Contradiction:
Improvepressure distribution measurement accuracyVSAvoidcalibration time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing calibration only once during manufacturing, establishing a correction matrix that is then stored and applied to all subsequent measurements. This eliminates the need for repeated time-consuming calibration procedures during normal operation, while maintaining high measurement precision through the pre-computed correction factors.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by creating a digital correction matrix that replicates the calibration results for all pixels. Instead of physically recalibrating each pixel during every measurement, the system copies the calibration data and applies it through matrix multiplication, significantly reducing calibration time while preserving accuracy.

Inventive Principle:
Principle #26Copying

2Measurement precision

If pixels are calibrated individually with separate force application, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improvepixel response accuracyVSAvoidcalibration process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges the calibration process by applying force simultaneously to all pixels using a uniform pressure source, rather than calibrating each pixel separately. This combined approach reduces calibration complexity while maintaining precision through the use of a correction matrix that accounts for individual pixel variations captured during the unified calibration process.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent implements universality by creating a single correction matrix that serves all pixels in the array. This universal correction approach simplifies the calibration process compared to individual pixel calibration, while still achieving high measurement precision by capturing the collective response characteristics of all pixels under uniform pressure.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If a mathematical model with algebraic correction laws is used, then measurement precision is improved, but ease of manufacture decreases

Engineering Contradiction:
Improvepressure distribution correction accuracyVSAvoidcalibration process simplicity
Core Design Contradiction:
Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent replaces complex algebraic correction laws with a computational approach using matrix operations. Instead of deriving and implementing complicated analytical correction formulas, the system uses a correction matrix computed from calibration data, which is applied through simple matrix multiplication. This substitution simplifies manufacturing by eliminating the need to determine and program complex algebraic relationships.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent applies parameter changes by transforming the calibration problem from finding complex algebraic correction laws to computing a correction matrix with numerical values. This parameter transformation simplifies the manufacturing process by replacing symbolic manipulation with numerical computation, making the calibration more straightforward while maintaining high correction accuracy.

Inventive Principle:
Principle #35Parameter changes

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The solution enables reliable measurement of pressure distribution, including non-trivial support forms, and simplifies the calibration process, making it more efficient and adaptable to different pixel distributions and configurations.

Implementation Method 1

a neural network for processing an image of the sensor response and providing a corrected image, this network having been trained from an augmented database

Methodology Applied
Scientific EffectNeural network processing:

Data Source

PatentEP4123281B1Matrix pressure sensor with neural network and calibration method
Publication Date: 2025.04.23 COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
  • EP4123281B1 patent drawingFigure 1~2B
  • EP4123281B1 patent drawingFigure 2C~2D
  • EP4123281B1 patent drawingFigure 3

AI summary

Neural network matrix pressure sensor and calibration method Matrix pressure sensor (1), comprising: - a matrix (2) of touch pixels (10) of which at least some exhibit a reciprocal coupling effect, - a neural network (30) to process an image (IP_MES) of the sensor response and provide a corrected image (IP_COR), this network having been trained from an augmented database (BDAUG) comprising: o real homogeneous support data measured by applying homogeneous pressure (PR) to at least a part of the pixels, better to all the pixels of the matrix and o additional partial support data produced by simulation by applying binary masks (MAS) to the real homogeneous support data, so as to simulate partial supports without coupling effect with the pixels located outside the partial support areas.