Stylus Trajectory Prediction With Polar Neural Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stylus trajectory prediction techniques suffer from low accuracy, particularly in angular errors, and high complexity, especially when hardware and software are sourced from different manufacturers, leading to poor user experience due to high touch-to-photon latency.
Innovation Solution
A neural network-based technique that minimizes angular error over other errors, using input normalization, polar coordinate conversion, and independent prediction of angular and radial components to reduce complexity and improve accuracy across various devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional stylus trajectory prediction techniques are used, then device complexity is reduced, but angular error increases leading to lower prediction accuracy
Solution Approach 1:
The patent divides the trajectory prediction problem into two independent segmentation components: angular prediction and radial prediction. The neural network separately predicts angular displacement and radial displacement, then combines them to generate the final trajectory point. This segmentation allows each component to be optimized independently, improving angular error accuracy while managing system complexity through modular architecture.
Solution Approach 2:
The patent transforms the traditional Cartesian coordinate prediction approach into a polar coordinate system approach. Instead of predicting x and y coordinates directly, the system predicts angular displacement and radial displacement in polar coordinates, then converts back to Cartesian coordinates. This dimensional transformation enables better angular error minimization by treating angular and radial components as independent prediction dimensions.
2Loss of time
If stylus trajectory prediction is implemented, then touch-to-photon latency is reduced, but implementation complexity increases especially across different manufacturers' hardware and software
Solution Approach 1:
The patent designs a universal neural network-based prediction system that can be implemented across different manufacturers' hardware and software platforms. The angular-radial prediction architecture provides a standardized interface that works independently of specific touch screen technologies (capacitive, resistive) or display systems, enabling broad compatibility while reducing touch-to-photon latency across diverse device ecosystems.
Solution Approach 2:
The patent replaces traditional physics-based or rule-based trajectory prediction mechanisms with a neural network-based predictive model. This substitution eliminates the need for manufacturer-specific calibration and tuning of mechanical prediction parameters, providing a universal software-based solution that reduces latency across different hardware configurations without increasing implementation complexity.
3Ease of operation
If minimal angular error prediction is prioritized, then user experience improves, but overall prediction complexity increases
Solution Approach 1:
The patent segments the prediction algorithm into distinct angular and radial prediction modules, allowing the angular prediction component to be optimized specifically for minimizing angular error and improving user experience. This segmentation enables the angular module to focus computational resources on angular accuracy while the radial module handles distance prediction, reducing overall algorithmic complexity compared to a monolithic approach.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
System and techniques for predicting a stylus position while interacting on a surface are described herein. The prediction begins by obtaining a set of points that are derived from a stylus moving on a surface. An artificial neural network (ANN) may be invoked on an input set. Here, the input set is based on the set of points from the stylus. The ANN is configured to output a next point from the input set, and the ANN is trained to minimize angular error for the next point over other errors. Once the next point is provided by the ANN, the next point may be communicated for rendering on a display.