Question

In an Extended Kalman Filter, what is the primary mathematical reason for linearizing a non-linear motion model using a Jacobian matrix?

Accepted Answer

The primary mathematical reason for using a Jacobian matrix in an Extended Kalman Filter is to enable the use of linear algebra tools for non-linear systems. The standard Kalman Filter requires motion and measurement models to be linear, meaning they must follow the form of a straight line where the output is directly proportional to the input. Most real-world motion models are non-linear, involving functions like sines, cosines, or squares, which prevent the direct application of the Kalman filter&#x27;s core equations. A Jacobian matrix is a matrix of all first-order partial derivatives of a vector-valued function. It acts as a local linear approximation of the non-linear function at a specific point, essentially creating a tangent line that represents the function&#x27;s slope in a multi-dimensional space. By calculating this Jacobian at the current estimated state, the Extended Kalman Filter converts the complex non-linear curve into a temporary linear model. This transformation allows the filter to calculate the state covariance, which represents the uncertainty of the estimate, using standard matrix operations like addition and multiplication. Without this linearization, the non-linear transformation would cause the probability distribution of the state estimate to become non-Gaussian, making it impossible to solve using the closed-form equations of the Kalman filter.

Home → All Courses → Engineering and Technology Courses → Autonomous Vehicle Engineering → Flashcard

In an Extended Kalman Filter, what is the primary mathematical reason for linearizing a non-linear motion model using a Jacobian matrix?