1. Human-Computer Interaction Kalman Filter Hanyang University Jong-Il Park

2. Rudolf Emil Kalman Born 1930 in Hungary Born 1930 iHungary BS and MS from MIT BS and MS from MIT PhD 1957 from Columbia Filter developed in 1960-61 Filter Now retired Now retired

3. What is a Kalman filter? Just some applied math. A linear system: f(a+b) = f(a) + f(b). Noisy data in hopefully less noisy out. But delay is the price for filtering... Pure KF does not even adapt to the data.

4. What is it used for? Tracking missiles Tracking missiles Tracking heads/hands/drumsticks Extracting lip motion from video Fitting Bezier patches to point data Lots of computer vision applications Economics Navigation Navigation

5. Typical Kalman filter application system measuring device Kalman filter measurement noise system noise system state (desired but not known ) controls observed measurement optimal estimate (system state)

6. Simple example

7. Observation 1

8. Observation 2

9. Combine the two Combine estimates Combine variances

10. Combined estimates Online weighted average!

11. But suppose we are moving … Not all the difference is error Some may be motion Some may be motion KF can include a motion model Estimate velocity and position

12. Process model Describes how the state changes over time The state for the first example was scalar The process model was nothing changes A better model might be A State is a 2-vector [ position, velocity ] position n+1 = position n + velocity n * time velocity n+1 = velocity n

13. Measurement model “ What you see from where you are ”

14. Predict  Correct KF operates by Predicting the new state and its uncertainty Correcting with the new measurement Correcting with the new measurement Predict Correct Kalman filter

15. Formulation

16. Kalman Filter

17. Kalman filter assumes linearity Only matrix operations allowed Measurement is a linear function of state Next state is linear function of previous state state If nonlinear  No way? projection

18. If nonlinear  Extended Kalman ilter(EKF) Nonlinear Process (Model)  Process dynamics: A becomes a(x)  Measurement: H becomes h(x) Filter Reformulation  Use functions instead of matrices  Use Jacobians to project forward, and to relate measurement to state

19. Formulation - EKF

20. Extended Kalman filter

21. Jacobian Partial derivative of measurement with respect to state If measurement is a vector of length M and state has length N  Jacobian of measurement fucntion will be MxN matrix of numbers (not equations) Often evaluating h(x) and Jacobian(h(x)) at the same time cost only a little extra

22. Simulation – Exact R

23. Convergence of covariance P

24. Simulation – Bigger R

25. Simulation – Smaller R

26. Concluding remarks Kalman filter is a very useful and powerful Optimal in many sense Linear system -> Kalman filter Nonlinear system -> Extended Kalman filter Effect of initial value  Covariance matrix – less sensitive  Measurement noise  Over-estimated -> slow convergence  Under-estimated -> noisy output

27. Resources Kalman filter website maintained by Greg Welch