Kalman Filter for Sensor Fusion

IMU-Based Orientation Estimation 📐

This repository contains Python scripts to estimate the 3D orientation (Yaw, Pitch, Roll) of a device using gyroscope and accelerometer data, and compares the performance of different estimation methods.

The pipelines are as follows:

• Euler Angle Integration (gyroscope only)
• Quaternion Kalman Filter (gyro only)
• Quaternion Kalman Filter (gyro + accelerometer)

---

🔧 Implemented Methods

Method	Description
Euler Angle Integration (Gyro)	Uses nonlinear Euler equations to integrate angular velocity (p, q, r) into Euler angles. Frame-aware but suffers from drift.
Quaternion Kalman Filter (Gyro Only)	Predict-only Kalman filter using quaternion propagation and gyroscope input. No correction from accelerometer.
Quaternion Kalman Filter (Gyro + Accel)	Full Kalman filter using gyroscope for prediction and accelerometer for orientation correction via quaternion update.

---

📈 Results

1. Euler Angle Integration from Gyroscope (Nonlinear)

Nonlinear frame-aware integration of Euler angles using only angular velocity. Simple and fast but can drift over time without correction.

---

2. Quaternion Kalman Filter (Gyro + Acceleration)

Full sensor fusion using gyroscope for predicting orientation and accelerometer for correcting pitch and roll drift. Stable and accurate.

---

3. Quaternion Kalman Filter (Gyro Only)

Performs prediction using quaternion kinematics and gyroscope readings but no accelerometer correction, so orientation drifts slowly over time.

🚀 How to Run

Install dependencies:

pip install numpy pandas matplotlib

Run each script to generate results and plots:

python euler_gyro_integration.py
python quaternion_kalman_gyro_only.py
python kalman_filter_gyro_accel.py

---

🧠 Author

Developed by Megha Kalia Inspired by the great lecture series by Dr. Shane Ross