Skip to content

1Higgin/Predictive_Scanning_Keyboard

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 

Repository files navigation

Predictive_Scanning_Keyboard

A one-handed, temporal input system designed for fast, intuitive text entry using predictive scanning and context-aware phrase selection.

Predictive Scanning Keyboard

Overview

This project explores a new approach to text input that replaces traditional spatial typing with sequential scanning and prediction.

Instead of pressing individual keys, the user selects from a dynamically ordered stream of predicted words and phrases. The system prioritizes likely continuations based on context, reducing the number of selections required to construct full sentences.


Core Principles

Temporal Input (Not Spatial)

  • Input is based on timing, not position.
  • Operates with a single input action (button, touch, or dwell).
  • Enables efficient one-handed use.

Prediction-First Design

  • Focus on likely next inputs, not raw characters.
  • Primary interaction is phrase-level, not letter-level.
  • Letters and symbols are fallback only.

Context-Aware Continuation

  • Tracks recent input (last 1–3 words).
  • Predicts:
    • next words
    • common phrases
    • grammatical connectors

Example: Input: I want Suggestions: to, a, the, to build


Phrase-Based Input

  • Outputs multi-word chunks (e.g. "want to", "build a").
  • Reduces total selections required.
  • Matches natural language thinking.

Adaptive Scanning

  • No fixed scan order.
  • Predictions are:
    • shown earlier
    • displayed larger
    • held longer (dwell bias)
  • Less relevant options appear later.

Mode-Free Interaction

  • No manual switching between:
    • letters
    • words
    • functions
  • Context determines what is shown.
  • Prediction replaces modes.

Cognitive Ergonomics

  • Slower scanning, but fewer decisions.
  • Visual hierarchy:
    • current item (highlighted)
    • next predicted item (emphasized)
  • Designed for anticipation and reduced mental load.

System Architecture

Input

  • Single or minimal input source
    • button
    • capacitive touch
    • dwell detection

Processing

  • Context tracking
  • Prediction engine (rule-based or adaptive)
  • Dynamic scan ordering

Output

  • Text generation (phrases, words, characters)
  • Optional: HID keyboard output

Key Differentiators

  • Temporal input vs spatial typing
  • Phrase prediction vs character entry
  • Adaptive scanning vs fixed layouts
  • Implicit modes vs manual switching
  • Speed achieved by reducing choices, not increasing input rate

Goal

To enable fast, low-effort sentence construction by aligning input with natural language flow, rather than traditional keyboard structures.


Status

Prototype stage:

  • Python (Tkinter) scanning interface
  • Context-based prediction in development
  • Hardware integration (touch sensors / microcontrollers) planned

Future Work

  • Predictive phrase expansion
  • Adaptive learning (user-specific patterns)
  • Hardware integration (capacitive touch / embedded systems)
  • HID keyboard implementation
  • UI refinement and visualization improvements

License

TBD

======

Physical Interface Concept

The system is designed to operate on a minimal, keyless touch surface, rather than a traditional mechanical keyboard.

Key Characteristics

  • No physical keys
  • Flat or slightly contoured surface
  • Input via:
    • capacitive touch
    • pressure / force sensing (optional)
    • single or multiple touch zones
  • Can be implemented as:
    • standalone device
    • embedded surface (e.g. desk, armrest, wearable)
    • hybrid with existing hardware

Why Remove Physical Keys

Traditional keyboards:

  • Require precise spatial targeting
  • Depend on finger positioning and travel
  • Scale poorly for one-handed use
  • Introduce mechanical complexity

This system replaces spatial accuracy with:

  • timing-based selection
  • predictive ordering
  • context-aware input

Advantages

Reduced Motor Demand

  • No need to locate or press specific keys
  • Works with minimal movement or a single contact point

Hardware Simplicity

  • Fewer moving parts
  • Lower mechanical wear
  • Potentially lower manufacturing complexity

Flexible Form Factor

  • Can be:
    • compact
    • wearable
    • integrated into other surfaces
  • Not constrained by key layout geometry

Scalable Input Surface

  • Same system works with:
    • one touch point
    • multiple zones
    • larger touch panels

Alignment with Predictive Input

  • Physical interface does not need to encode meaning
  • All complexity handled in software (prediction + scanning)

Trade-Offs / Limitations

Lack of Tactile Feedback

  • No physical key press confirmation
  • May require visual or haptic feedback alternatives

Learning Curve

  • Users must adapt to:
    • scanning timing
    • predictive selection
  • Different from conventional typing habits

Dependence on Prediction Quality

  • System efficiency relies heavily on:
    • accuracy of predictions
    • relevance of suggested phrases

Slower Raw Input (Without Prediction)

  • If prediction fails, fallback (letters/functions) is slower than a keyboard

Visual Dependency

  • Requires user attention on display
  • Not suitable for fully blind typing without additional feedback systems

Design Position

This is not a direct replacement for a mechanical keyboard in all contexts.

Instead, it represents:

  • a software-driven input layer
  • decoupled from physical key constraints
  • optimized for low-effort, predictive sentence construction

The physical surface is intentionally simplified so that:

intelligence is moved from hardware into the interaction model

About

A one-handed, temporal input system designed for fast, intuitive text entry using predictive scanning and context-aware phrase selection.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors