CarND-Extended-Kalman-Filter-Project

In this project a Sensor Fusion Model is implemented in C++ to fuse LIDAR and RADAR sensor data into an Extended Kalman Filter for Localization of a moving car.

Kalman Filter is a loop of 2 Major Steps:

1. Predict - Calculated State Vector x_ and State Covariance P_
2. Update - Incorporate the new sensor data to the predicted state, and correcting the x_ and P_.

These Following Steps are Executed in order to Calculate x_ and P_ upon a sensor measurment:

1. Determine dt to further calculate the F and Q matrices.

2. Predict State Vector x to determine x: position, velocity - P: State Covariance

   • LIDAR and RADAR prediction functions are the same.

3. Determine whether the detected measurment is from a RADAR or LIDAR sensor:

   1. RADAR - Using Extended Kalman Filter Equations:
   • Convert Input Polar Coordinates to Cartesian using the h function to calculate Error y.
   • Calculate Jacobian Matrix H to calculate the Kalman Gain K while maintaining linearity, including the R measurement noise covariance matrix.
   • Use calculated y and K to correct for the predicted x and P.
   2. LIDAR - Using Standard Kalman Filter Equations:
   • Directly Calculate the Error y
   • Calculate the Kalman Gain Matrix K, including the R measurement noise covariance matrix.
   • Use calculated y and K to correct for the predicted x and P.

---

*For More Information check the source files to review the EKF algorithm.