Sensor-Fusion-and-Nonlinear-Filtering

Repository for the course "Sensor Fusion and Nonlinear Filtering"

Final Assignment: Orientation estimation of a cellphone using EKF and Sensor Fusion

In this project accelerometer, magnetometer and gyroscope measurements were used in a sensor fusion fashion to estimate the orientation of a cellphone.

This project includes:

• Derivation of the motion model using quaternions representation
• Derivation of measurement models
• Implementation of EKF subject to outlier rejection and sensor failure
• Detailed analysis of the implemented EKF

Assignment 1: Gaussian distribution analysis

• Transformation of Gaussian random variables
• Analysis of joint distribution and posterior densities
• MMSE and MAP estimates for Gaussian mixture posteriors

Assingment 2: Linear Kalman filter analysis, Motion and Measurement models

• A first Kalman filter and its properties
• Study of different Motion and Measurement models
• Kalman filter and its tuning

Assignment 3: Nonlinear Kalman filter models (EKF, UKF, CKF)

• Approximations of mean and covariance
• Non-linear Kalman filtering
• Tuning non-linear filters

Assignment 4: Smoothing, Particle Filter and Localization

• Non-linear RTS smoothers
• Particle filters for linear and Gaussian systems
• Localization: bicycle tracking in a village using Particle Filters