Inertial Navigation Unit

Legacy context

The Applied Analytics heritage in this domain traces directly to the Position and Attitude Data System (PADS) and the GPS Inertial Data Simulator (GIDS). PADS established a benchmark for high-accuracy, near real-time airborne direct georeferencing by tightly integrating a Kearfott IMU with GPS observations through a Multi-State Kalman filter. That system delivered a complete inertial navigation solution at the IMU data rate, with precise UTC event timing and in-air alignment capabilities. GIDS extended that expertise into simulation, providing a hardware-based method to generate real-time GPS and IMU data streams for bench-top checkout of systems with embedded inertial navigation units.

That documented background in strapdown solutions and sensor fusion is directly relevant to the modern inquiry around the inertial navigation unit. The core challenge remains consistent: maintaining accurate position and attitude when GPS is unavailable or degraded. The legacy work on Kalman-filtered alignment and high-rate data processing informs how contemporary units are evaluated for drift, initialization, and re-start behavior. This site’s history is not about abstract theory; it is about the practical integration of inertial sensors with external aiding sources. That foundation provides a useful context for exploring the specifications, interface requirements, and performance trade-offs of current inertial navigation units in the field.

Core Architecture and Sensor Characteristics

An inertial navigation unit (INU) forms the self-contained core of an integrated navigation system, detecting instantaneous vehicle linear acceleration along three orthogonal axes and deriving linear velocity, position, attitude, and heading from those measurements [4]. The sensing elements typically comprise a vertical accelerometer, two horizontal accelerometers, and gyroscopes mounted in a gimbal-stabilized platform, with attitude and heading information obtained from resolver devices mounted between the platform gimbals [4]. A representative high-accuracy system measures 9.24 inches high, 8.49 inches wide, and 22 inches long, weighing 43.5 pounds [4]. These physical dimensions and mass figures provide engineers with a baseline for platform sizing when integrating an INU into a vehicle architecture.

The inertial measurement units provide integrated accelerometer and gyro data at high rate, with the inertial state propagated at 40 Hz and updated by GPS pseudorange and deltarange measurements at 1 Hz in one flight-proven architecture [6]. This rate separation is a practical design choice: the inertial propagator runs fast enough to capture vehicle dynamics, while the GPS update rate is sufficient to bound the inertial error growth. In another implementation, navigation equations use an Euler second-order integration method at 100 Hz, chosen to match the technique typically used in inertial-only systems [7]. The selection of integration rate depends on the vehicle dynamics and the accuracy required; higher rates reduce integration error but increase computational load.

Error Sources and the Rationale for Integration

The fundamental limitation of an inertial navigation unit is that uncorrected sensor errors integrate over time, leading to unbounded position and velocity errors. Accelerometer and gyroscope bias terms must be included in the filter dynamics, particularly for long coast periods on orbit, because any uncorrected sensor errors continue to integrate over hour-long trajectories, leading to increased state dispersions [7]. As shown in trajectory analyses, integration errors grow to large levels over in-space trajectories, and GPS measurements support estimation of these error terms, minimizing integration errors due to sensor uncertainties [7].

The integration of INS and GPS addresses this weakness by combining complementary characteristics. The INS provides high-rate, self-contained measurements that are not dependent on external signals, while GPS provides absolute position updates that bound the inertial drift. The main question addressed in integrated navigation design is what benefits arise from the integration of INS and GPS and how this navigation concept is achieved in reality [2]. The answer lies in the error estimation architecture: filters play a central role in optimal estimation of navigation errors, and the discussion of error types and sources in navigation, and of the role of filters in optimal estimation of the errors, forms the theoretical foundation of integrated systems [2].

Filter Architectures and Calibration Considerations

A common integration approach uses a two-filter structure: a dynamics filter that uses GPS carrier-phase measurements to estimate velocity and other IMU errors, and a position filter that uses the velocity output of the dynamics filter and GPS pseudorange measurements [5]. This separation of velocity and position estimation allows each filter to operate at a rate appropriate to its measurements and reduces the computational burden of a single large filter. The dynamics filter runs at the GPS measurement rate, while the position filter can run at a lower rate since position changes more slowly than velocity errors accumulate.

Calibration of the inertial platform is a critical operational task. Calibration software improves the calibration of the inertial platform measuring unit gyroscopes and accelerometers, and the navigation program may be implemented as a wander axis system where the heading axis is not torqued [3]. This wander axis implementation frees the system from azimuth torquer errors, but requires that azimuth gyro drift be measured and used to calculate platform heading [3]. The benefit is that platform alignment for navigational use is quicker because the cluster does not have to be physically aligned to a known heading [3]. Engineers should note that the wander axis approach trades mechanical complexity for computational complexity and requires accurate gyro drift estimation.

Practical Design Guidance

When designing an integrated INS/GPS system, engineers should consider the rate at which the inertial state must be propagated. The 40 Hz propagator with 1 Hz GPS updates [6] and the 100 Hz Euler integration [7] represent two validated approaches, but the appropriate rate depends on the vehicle's dynamic environment. A slowly maneuvering spacecraft may require only 40 Hz, while a more agile platform may need the higher rate to capture rotational dynamics accurately.

The physical configuration of the inertial platform also matters. The four-gimbal, gyro-stabilized platform maintains accelerometers in a known reference frame through gyroscope control, with the accelerometers as the primary source of information [4]. This mechanical stabilization approach differs from strapdown systems, where the accelerometers and gyroscopes are fixed to the vehicle body and the navigation solution is computed in software. The choice between gimbaled and strapdown architectures involves trade-offs in mechanical complexity, reliability, and computational requirements.

For long-duration missions, the inclusion of accelerometer and gyroscope bias states in the filter is essential [7]. Without bias estimation, even small sensor errors will accumulate over hours of operation, producing unacceptable position errors. The filter must also account for the rotation between body and inertial frames when including these error terms in the dynamics [7]. This coupling between attitude and translation errors is a key consideration in filter design, as attitude errors manifest as acceleration errors through the rotation matrix.

The integration of GPS and INS provides fault tolerance and long life compared to either system operating alone [2]. The INS continues to provide navigation during GPS outages, while GPS prevents the unbounded growth of inertial errors during extended operations. This complementary operation is the primary benefit of integrated navigation and the reason it has become the standard approach for aerospace vehicles requiring accurate, reliable position and attitude information.