ESA GNC Conference Papers Repository

On-board real-time gyro calibration is an enabling technology for future missions requiring high-precision and fast target acquisition and tracking. This paper tackles the major challenges to be mastered, provides latest solutions and achieved performances. The paper focus is not on academic exercised but on practical needs and solutions for application on board spacecraft. This is the purpose of the Gyro Calibration Filter Library (GCFLib) developed at Airbus Defence & Space, Germany. Motivation As indicated below and detailed in the paper, the need for gyro calibration is motivated by the need for spacecraft attitude determination. It is achieved by attitude and gyro measurement fusion. It can generally be expected that many future Earth observation or science spacecraft, interplanetary science probes, and exploration robots will require a high-precision attitude determination and a high level of autonomy, reliability, and availability. Furthermore, this has to be achieved on short procurement schedules, at low cost and low risk. Achieving these goals only through improvements in attitude and angular rate sensors is impractical, especially given the high cost, long development and procurement cycles of high performance hardware and related risk. As consequence, on-board real-time gyro calibration algorithms are necessary to improve system performance and enable future space missions. Gyro calibration, i.e. determination of gyro biases, scale factor errors, misalignments, is especially important for spacecraft that perform slews to acquire stationary targets such as stars or to track targets. Accurate target acquisition is especially required to reduce acquisition time and, thus, to increase mission efficiency. It is also very important for accurate attitude control during attitude sensor drop-outs. Sticking to a conventional 6-state gyro-stellar estimator instead can lead to disastrous consequences or even loss of mission. Estimator approach The paper provides a systematic trade and justification of the most suitable estimation filter for on-board real-time gyro calibration. It considers estimators such as linear, extended and unscented Kalman filter and different approaches to account for unknown gyro calibration parameter errors. It will be shown whether conventional body-frame parameters or physical parameters should be estimated. And which solutions can handle redundant gyro units such as the Astrix-200 or SIRU on ESA Sentinel-2 or BepiColombo, respectively. The resulting approach in form of equations will be presented. Observability issues Often underestimated but of major importance in gyro calibration filters are the observability aspects. Full observability is needed to estimate all gyro parameters modelled in the calibration filter and thus to enable fast high-precision repointing of spacecraft. However, full observability is only achieved if both geometric and dynamic observability is achieved. While the first is driven by the number of gyros, their alignment and a specific handling in the estimator, the latter is treated by specific rotational calibration manoeuvers. The paper will detail these aspects. The ‘observability dilemma’ A common wish during spacecraft design and development is to mount the gyro unit remote from attitude sensors or payload with the goal of reducing engineering design effort and cost, and possibly relaxing operational constraints. Typically, such a solution would cause temperature-induced gyro parameter variations. How, to what extend and at what cost such time-variations can be considered will be presented in the paper. Benchmark results Three different estimator solutions to handle gyro parameter knowledge errors will be compared: (1) conventional 6-state attitude estimator without slew-dependent covariance tuning, (2) like (1) but with slew-dependent covariance tuning, and (3) the proposed gyro calibration filter. Benchmark is a slew manoeuvre with fast high-precision repointing.