🕺 MoveSpace Processor (MSP)

MoveSpace Processor (MSP) is a cross-platform motion analysis engine designed to evaluate human movement data—such as accelerometer and gyroscope readings—against pre-trained motion models stored in .msm files. Inspired by Ubisoft's Just Dance MARS system, MSP enables real-time motion scoring and classification, making it ideal for motion-driven games, fitness tracking, and movement quality assessment tools.

---

🏗 Project Structure

MSP/
├── javascript/       # JavaScript & TypeScript implementation (reference version)
├── python/           # Python port (completed)
├── csharp/           # C# port (planned)
├── c++/              # C++ port (planned)
├── example/          # Example .msm models and test data
├── README.md         # Project documentation

---

🚀 Features

• ✅ Cross-platform motion scoring engine
• ✅ Reads and decodes binary .msm files
• ✅ Analyzes acceleration vector data
• ✅ FFT, RMS, direction, and stability metrics
• ✅ Supports Mahalanobis and Naive Bayes classifiers
• ✅ Modular architecture for extensibility
• ✅ Implemented in JavaScript/TypeScript and Python

---

🧪 Sample Use Case (JavaScript)

import { Model, Analyzer } from '../javascript/msp.js';

// Step 1: Load a MoveSpace model (.msm) as binary data
const buffer = await fetch('./generic_generic.msm').then(r => r.arrayBuffer());
const model = new Model(buffer);

// Step 2: Initialize the analyzer with one or more models
const analyzer = new Analyzer([model]);

// Step 3: Feed motion samples into the analyzer
// ⚠️ Timestamps must be in milliseconds, matching real-time intervals
// Acceleration values should be in raw sensor format [x, y, z] (unit: m/s² or Gs depending on source)
analyzer.addSamples([
  { timestamp: 0, accel: [0.17, -0.79, 0.34] },      // t = 0 ms
  { timestamp: 20, accel: [-0.45, -1.04, -1.32] },   // t = 20 ms
  { timestamp: 40, accel: [-1.47, 0.71, 0.59] },     // t = 40 ms
  // ...
]);

// Step 4: Analyze a motion segment
// - 'model.modelName' should match the model name inside the .msm file (e.g., 'generic')
// - '0' is the start time in milliseconds (relative to the first sample)
const result = analyzer.analyzeMove(model.modelName, 0);

// Step 5: View the results
console.log("Composite Score (0-100):", result.score);
console.log("Breakdown:", result.components);

/*
Output:
Composite Score (0-100): 72
Breakdown: {
  statistical: 0.729,
  autoCorrelation: 1,
  direction: 0,
  stability: 0,
  classifier: 0
}
*/

---

📊 Simulated Output

Using random motion sample (50 vectors) compared to generic_generic.msm

Metric	Value
Model Name	generic
Map Name	Generic
Measure Set	Acc_Dev_Dir_NP
Model Duration	652 ms
Classifier Type	Naive Bayes
Reference Mean	~4.90e-44 (zero)
RMS Error	1.75
Normalized Score	0.729

---

🧠 What’s Inside a .msm File?

• Endianness marker
• Version
• Model, Map, and Measure Names
• Duration (in seconds)
• Low/High Thresholds
• Auto-correlation Thresholds
• Direction Impact
• Classifier Block (means, covariance, energy)

---

📘 References

• Ubisoft Just Dance's Mars System (original inspiration)
• Dynamic Time Warping
• Mahalanobis Distance
• Cosine Similarity
• Fast Fourier Transform

---

🔄 Language Support

Language	Status	Notes
JavaScript/TypeScript	✅ Completed	Reference implementation
Python	✅ Completed	Matching logic and I/O
C++	🔜 Planned	High-performance runtime
C#	🔜 Planned	Ideal for Unity integration

---

📜 License

This project is part of the LilyPad Next Licensed under the Apache License 2.0