Kalman Filter in practice
This part contains practical guidelines for Kalman Filter implementation:
- Sensors Fusion
- Variable measurement error
- Treating missing measurements
- Treating outliers
- Kalman Filter Initialization
- KF Development Process
Sensors Fusion
Many practical systems are equipped with several complementary and sometimes interchangeable sensors that measure the same parameters.
A self-driving car has Light Detection and Ranging (LiDAR) and and radar onboard. The LiDAR is much more precise than the radar. On the other hand, the radar measures velocity using the Doppler Effect, and its effective operational range is higher, especially in rain or fog conditions. The aircraft is equipped with Global Navigation Satellite System (GNSS) and Inertial Navigation System (INS) syste systems for navigation. Many surveillance systems include several radars for target tracking.
Using multiple sensors can significantly improve the state estimation precision in a process known as sensor fusion.
Sensor fusion refers to combining the measurements from multiple sensors resulting in joint information having less uncertainty than any of the sensors individually.
