According to Phil Kim, understanding a few basics is more important than complex math: The true variable you want to know (e.g., location). Measurement ( The noisy data received from a sensor. Estimation Error Covariance ( cap P sub k How uncertain the filter is about its estimate. Process Noise Covariance ( How uncertain the system model is. Measurement Noise Covariance ( How noisy the sensor is. DSPRelated.com 3. The 5-Step Kalman Filter Algorithm The filter operates in a loop: Prediction (Time Update) Project the State Ahead: Estimate the next state based on the current state. Project the Error Covariance Ahead: Predict how uncertainty grows. Update (Measurement Update) Compute Kalman Gain ( cap K sub k
% Update K = P_pred / (P_pred + R); x = x_pred + K * (measurements(i) - x_pred); P = (1 - K) * P_pred; According to Phil Kim, understanding a few basics