Kalman filter
- In French:
- Filtre de Kalman
Definition
An estimation algorithm that combines a prediction based on a model of the motion with imperfect measurements to estimate the state of a system, for example a robot’s position and speed. It alternates two steps: predicting the next state, then correcting it with the new measurement, taking into account the uncertainty of each. Its extended version (EKF) handles non-linear systems; the ROS package robot_localization uses it to fuse sensors.
Checked on against the references below.