Implementation of Unscented Kalman Filter_Vision+IMU

The goal of this project is to implement sensor fusion using the Unscented Kalman filter by making use of the IMU data and the vision-based pose and velocities estimated. The vision-based position and orientation is obtained through projective transformation and the linear and angular velocity are calculated through optical flow. Since this system is highly non-linear, the Unscented Kalman filter captures the non-linearity of the model much better to estimate the state of the quadrotor moving in an experimental space.

System State:

Refer documentation for detailed explanation of implementation

Execute this implementation:

• Clone this repository and open the directory within the MATLAB environment. Run UKF_kalmanfilt_Part1.m to visualize outcomes of UKF using Camera for measurement(apriltag pose estimation)
• Run the UKF_kalmanfilt_Part1.m to understand the outcomes of UKF using velocity estimated using optical flow and how the approach captures the non-linearity of the system.