Pressure Input Classification in Sensing Arrays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Force sensing touch screens face challenges in accurately classifying pressure inputs, particularly when users apply additional pressure with their palms, leading to undesirable activations and errors in pressure calculations.
Innovation Solution
A method and apparatus for classifying pressure inputs in a sensing array, which involves identifying both desirable and undesirable pressure inputs, converting them into an output image, and using an artificial neural network to apply a mask and remove undesirable inputs, thereby isolating and enhancing the desirable inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the sensing array detects all pressure inputs, then the sensitivity and detection capability are improved, but false activations and calculation errors increase due to palm pressure
Solution Approach 1:
The patent segments the pressure input data by spatially dividing the sensing array into different regions (stylus contact region vs. palm contact region) and temporally analyzing pressure patterns. By segmenting the input signal into desirable and undesirable components based on location and temporal characteristics, the system can accurately detect all pressure inputs while distinguishing between valid stylus inputs and unwanted palm pressure.
Solution Approach 2:
The patent introduces an intermediary processing layer that analyzes the spatial and temporal characteristics of pressure inputs. This intermediary system uses machine learning models to mediate between the raw pressure data and the final interpretation, automatically filtering out palm pressure while preserving stylus inputs based on their distinct patterns without requiring manual configuration.
2Reliability
If the system applies pressure threshold filtering, then false activations from light palm pressure are reduced, but legitimate stylus inputs with varying pressure may be missed
Solution Approach 1:
The patent implements dynamic threshold adjustment where the pressure threshold is not fixed but adapts based on the detected pressure pattern, location, and temporal characteristics. The system dynamically modifies acceptance criteria for pressure inputs based on real-time analysis of input patterns, allowing it to accept legitimate stylus pressure variations while rejecting palm pressure across different force levels.
Solution Approach 2:
The patent changes multiple parameters simultaneously including pressure threshold, spatial location weights, and temporal duration criteria based on the detected input pattern. By adjusting these parameters dynamically according to the situation, the system maintains high reliability in rejecting false inputs while preserving adaptability to recognize legitimate stylus inputs across the full pressure range.
3Measurement precision
If machine learning models are trained extensively, then classification accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary training of machine learning models during device manufacturing or initial setup, so that the classification system is pre-configured with learned patterns for distinguishing stylus from palm pressure. This preliminary action transfers the computational burden from real-time operation to offline training, enabling fast real-time classification with minimal processing delay during actual use.
Solution Approach 2:
The patent replaces complex real-time computational analysis with simplified decision rules derived from pre-trained machine learning models. The system substitutes heavy online computation with lighter-weight inference operations that have been optimized during training, achieving high classification accuracy while minimizing processing time and computational resource requirements during actual touch detection.
Data Source
AI summary
A method of classifying pressure inputs in a sensing array, in which the sensing array comprises a plurality of sensing elements responsive to pressure inputs is described. The method comprises steps of identifying a plurality of pressure inputs in the sensing array and converting the plurality of pressure inputs into an output image. The output image is compared with a data set comprising a plurality of images of undesirable pressure inputs by means of an artificial neural network. A mask is applied which is consistent with the output image to remove undesirable pressure inputs.