Lecture 14: GNSS/INS Integration and Kalman Filtering

Vocabulary terms and critical concepts from Lecture 14 covering GNSS/INS integration architectures, timing errors, and initialization procedures.

Uncoupled Integration

The simplest method of coupling GNSS and INS where the inertial solution is periodically re-initialized based on external references, allowing the INS to run freely between resets.

Open Loop System

An architecture where corrections are applied in the solution domain, affecting Position, Velocity, and Attitude directly.

Closed Loop System

The standard modern architecture where corrections are applied in the INS measurement domain, such as Accelerometer and Gyroscope biases.

Loose Coupling

A common integration architecture where the GNSS receiver provides computed position ($r$) and velocity ($v$) to the integration filter.

Observability Issue (Loose Coupling)

A limitation where the integration gains no information from the GNSS receiver if fewer than 4 satellites are visible, as the receiver cannot calculate a position solution.

PPS (Pulse Per Second)

An electrical pulse signal from the GNSS receiver used to synchronize the internal free-running clock of the INS to the GNSS measurement epoch.

Solution Latency

The finite amount of time required by a receiver processor to perform calculations before data can be transmitted.

Communication Latency

The delay introduced by packing, moving, and unpacking data across a medium, such as an RS232 serial stream at $115200\,\text{baud}$.

Federated Filtering

An architecture that partition models into sensor-specific filters, providing performance benefits and simplifying the handling of additional error states like wheel speed sensors.

Tight Coupling

An integration method that works in the GNSS measurement domain using raw pseudoranges and range rates, allowing the system to utilize data from fewer than 4 satellites.

Deep Coupling

An advanced integration that feeds back to the GNSS receiver's correlator states (code phase and code phase rate) using digital spectrum samples (In-phase and Quadrature samples).

ZUPT (Zero Velocity Update)

A virtual measurement applied when a platform is detected as static to reset growth in the system velocity error state.

ZARU (Zero Angular Rate Update)

A requirement used alongside ZUPT to ensure the platform is non-rotating, often used in low-cost INS to estimate gyro biases.

Coarse Alignment

The initial estimation of roll ($\phi$), pitch ($\theta$), and yaw ($\psi$) using gravity for leveling and the earth rotation rate ($7.29 \times 10^{-5}\,\text{rad/sec}$) for heading.

$\phi$ (Phi)

The symbol used in mechanized navigation equations to represent the roll state.

$\theta$ (Theta)

The symbol used in mechanized navigation equations to represent the pitch state.

Pull-in Range

The specific error range within which a system's yaw estimate will eventually self-correct once the platform begins moving.

Scale Factor Error

A type of sensor error that, unlike bias, requires the magnitude of the observation (acceleration or rotation) to be modulated to become observable.

Virtual Measurements

Information added to the navigation system based on knowledge of platform dynamics, such as Zero Velocity Updates or motion constraints.

Van Loan

A reference to a researcher known for the paper 'Nineteen Dubious Ways to Compute the Exponential of a Matrix', relevant to matrix calculations in Kalman Filtering.