Ankit Vora

Phase-1: Implement a complimentary filter

The first phase of the project dealt with designing a complimentary filter to estimate the state of the quadrotor using the onboard IMU data. Complimentary filters are basically weighted average of estimates of the gyro and acceIometer readings. This is used as an alternate to using complex filters like Kalman filter. The figure shows the ground truth (red) and the filter output (blue).

Phase-2: Implement the Uncented Kalman Filter (UKF)

This phase dealt with designing an Unscented Kalman Filter using quaternion algebra. The filter parameters like noise and model co-variance were learnt from the dataset given using the cross validation technique. The figure here shows the filter output (blue) as compared to the ground truth (red).