Evolution Operator
Mathematical operators that define how a system’s state vector changes over a discrete time step are fundamental to predictive modeling. The state propagation matrix maps the current position, velocity, and orientation to their expected values at the next interval. It embodies the physical laws governing the system’s motion.
Matrix Composition
The elements of the matrix are derived from the differential equations of motion. In a navigation filter, the state propagation matrix includes terms for the time step and the relationships between derivatives. For a linear system, this matrix remains constant.
Computational Role
During the prediction phase of a filter, the state propagation matrix is multiplied by the current state estimate. This result represents the best guess of where the system will be before the next sensor measurement arrives. The same matrix is also used to project the state covariance forward in time.
This projection accounts for the uncertainty added by the system’s motion. If the dynamics are non-linear, the matrix is updated at each step using a Taylor series expansion. Accuracy in this matrix is necessary for maintaining a stable track between sensor updates.
Error Accumulation
Any inaccuracy in the underlying model within the state propagation matrix will lead to a growing divergence from the truth. High-fidelity systems use higher-order integration schemes to minimize this effect.
