Kalman Sensor Studio v4.1 - Universal Edition

Enhanced Signal Processing

New in v4.1

• Transfer Entropy Analysis: Shannon entropy, mutual information, directional TE for sensor coupling detection
• Adaptive Fuzzy Kalman Filter: Fuzzy innovation-based Q/R adaptation with 3 fuzzy rules, dynamic sigma scaling

Features

• Zero new dependencies (pure JavaScript)
• 4 new IPC handlers and UI panels

Kalman Sensor Studio v4.0.0

Kalman Sensor Studio v4.0.0

What is New

• Complete rewrite as Electron desktop app
• Pure JS Kalman filter: Scalar (1-State) + Constant Velocity (2-State)
• RTS Smoother for backward-pass smoothing
• Signal Generator with 7 types (sine, cosine, square, sawtooth, step, ramp, constant)
• Grid search auto-tuner with RMSE optimization
• Interactive Q/R sliders with real-time updates
• Uncertainty visualization with plus/minus 1-sigma bands
• Innovation sequence analysis
• CSV import/export via native dialogs
• Professional cyan-themed dark UI
• Zero runtime dependencies

Tech Stack

• Electron 28 (desktop shell)
• Pure JavaScript (zero npm dependencies for core)
• Chart.js 4.4.7 (visualization)
• electron-builder (Windows .exe packaging)

Build on Windows

npm install
npm start

For portable .exe:

npm run build-portable

Kalman Sensor Studio 1.0.0

Kalman Sensor Studio 1.0.0

First desktop release: a portable Windows application for real-time Kalman filtering, tuning and analysis of noisy sensor streams.

Download

Kalman-Sensor-Studio-win32-x64.zip -> unzip -> run Kalman Sensor Studio.exe. Portable, no installer, no runtime dependencies.

Highlights

• Live streaming workspace: sine / ramp / step / random-walk sources, adjustable noise, 5-120 Hz sample rate, outlier injection
• Two filter models: scalar constant-state and 2-state constant-velocity with full covariance propagation
• Interactive log-scale tuning of process noise q and measurement noise r
• Auto-tune: grid search minimising one-step-ahead prediction RMS error
• Optional RTS (Rauch-Tung-Striebel) offline smoother
• +/-1 sigma confidence band and a dedicated innovation (residual) chart
• Live metrics: RMSE vs ground truth (raw vs filtered), noise reduction, Kalman gain, covariance, sample count
• CSV/TSV import to replay real sensor logs; export filtered session to CSV and chart to PNG
• Keyboard shortcuts and native menus (Space, Ctrl+O, Ctrl+S, Ctrl+T, Ctrl+R)

Library

kalman.js now exports ScalarKalman, ConstantVelocityKalman, rtsSmooth, autoTune and metrics, still dependency-free. All tests pass with npm test.

MIT licensed.