Satellite Launcher Navigation with One Versus Three IMUs: Sensor Positioning and Data Fusion Model Analysis

Abstract

Using multiple IMUs allows both their distribution along vehicle structures and a reliance on integration methods, which is not possible with a single IMU. This paper addresses the issue of relying on three IMUs instead of only one of a higher quality in the context of a satellite launcher. The impact of the IMU positions was tested by comparing collocated IMUs against IMUs installed in the head of each launcher stage. For multi-IMU configurations, three integration methods were tested: all IMUs fused in a single INS, multiple INSs fused in a stacked filter, and multiple INSs fused in a stacked filter with geometrical constraints. All navigation solutions were aided by a three-axis attitude reference sensor and were tested with and without a GPS receiver. The results show that distributing IMUs along the launcher structure does not improve navigation performances compared to having them collocated. The fusion of multiple IMUs in one INS provides equivalent results as one IMU. However, fusing multiple INSs greatly reduces estimation errors. Performances are further improved with the addition of geometrical constraints. During long GPS outages, relative velocity and position constraints should not be exploited, as they may lead to filter divergence.

Multi-IMU Based Alternate Navigation Frameworks: Performance & Comparison for UAS

“…[2] used an ad-hoc Kalman filter integrating two IMUs for marine navigational purposes. Position estimation accuracy from multi-IMU approaches can significantly exceed the single IMU case, as indicated by [3], which compared satellite launcher position estimation and showed the three IMU case reduced error by 54% relative to the single IMU case. [4] applied the [3] framework to an integrated navigation system that includes a traditional inertial navigation system (INS) with auxiliary IMU sensors.…”

“…Other studies that are more closely related to this research have focused on dynamic systems analyzed all frameworks individually rather than an extensive framework comparison [16], [17]. Where as other researches focused on the orientation of sensor placement [18], [19], [20], [21]. Despite the work in this area to develop multi-IMU fusion approaches, experimental data for dynamic systems on their comparative performance remains sparse.…”

“…As a result, the question of which approaches are best implement in an airborne UAS platform requiring higher accuracy is not yet answered quantitatively for high rate dynamic systems. This work begins to answer that by systematically outlining and comparing four estimation strategies based on previous literature and most commonly referenced [3], [11], [15], [22]. The results include both theoretical and experimental results of the approaches implemented in an on-board UAS autopilot and on the same measurements, allowing direct comparison.…”