Next

Getting Started

The Nonlinear Estimation Toolbox relies on four basic concepts:

  1. Estimators (also called filters)

  2. Measurement models

  3. System models

  4. Probability distributions

To solve a given estimation problem with the toolbox, you have to:

  • choose the estimator that fits best to the problem (e.g., a Kalman filter or a particle filter) and configure it appropriately (e.g., set the number of particles),

  • choose the measurement model that fits best to the problem and implement the required functions (e.g., the measurement equation or the likelihood),

  • choose the system model that fits best to the problem and implement the required system function, and finally

  • choose the probability distributions imposed by the problem (e.g., a Gaussian mixture for the initial system state, and Gaussian distributions for the measurement and system noise, respectively).

Let's try it out with a first illustrative example!

The following example code can be found in the toolbox's examples.

A Linear Estimation Problem — Solve your first estimation problem with the toolbox
A Nonlinear Estimation Problem - Part I — Solve a nonlinear estimation problem with linear filters
A Nonlinear Estimation Problem - Part II — Solve a nonlinear estimation problem with nonlinear filters.
Simulate System State Trajectories and Measurements — Simulate system state trajectories and measurements based on the implemented system and measurement models.
Run a Set of Filters — Run and compare several filters the easy way.
Simulate a Complete State Estimation Problem — Combine the features of the Nonlinear Estimation Toolbox to evaluate several filters in a complete state estimation simulation.
Next