Laser range finder closed-loop pointing technology of relative navigation, attitude determination, pointing and tracking for spacecraft rendezvous

[0001] This application claims priority to and the benefit of the filing date of provisional application entitled “Laser Range Finder Closed-Loop Pointing Technology of Relative Navigation, Attitude Determination, Pointing and Tracking for Spacecraft Rendezvous,” assigned Ser. No. 60/492,596 and filed Aug. 5, 2003, which is hereby incorporated by reference.

[0002] The present invention relates generally to electronic control systems for rendezvous of spacecraft in orbit, and, more particularly, to laser range finder closed-loop pointing control systems of relative navigation, attitude determination, pointing, tracking, and guidance subsystems for autonomous spacecraft rendezvous.

[0003] Navigating and controlling a spacecraft for rendezvous with another spacecraft in a circular orbit is a difficult and complicated process. Further compounding this process are the precise nature and errors of the measurements for navigating and controlling a spacecraft for rendezvous with a spacecraft in a circular orbit.

[0004] Spacecraft navigation systems typically include multiple components. These systems may include various types of sensors to provide measurements for the component systems, such as visible and infrared sensors, laser range finders, mechanical or ring laser gyros, accelerometers, Sun sensors, star trackers, and GPS (global positioning system) receivers. These instruments often are misaligned and their measurements are corrupted with noise. Kalman filters may be used to provide estimates of relative position, velocity, attitude and rates.

[0005] When two spacecraft in near-Earth circular orbits are within a distance of approximately 10 km, guidance policies for a chaser satellite may be based on linear Clohessy-Wiltshire equations. The calculation of guidance delta-velocities utilized by the Clohessy-Wiltshire equiations for rendezvous and proximity operations requires position and velocity estimates of the chaser satellite relative to the target satellite.

[0006] For this process to be automated, a laser range finder (LRF) instrument, mounted on a chaser satellite, may be pointed to a target satellite so that the range between the two spacecraft may be measured. However, because of the narrow width of the laser beam and small target size, the laser range finger (LRF) must be pointed accurately at the target to approximately 90 μrad/axis, or 1σ, in both azimuth and elevation from a distance of approximately 10 km. This stringent pointing accuracy must be accomplished with a collimated visible sensor while multiple factors further complicate the process, such as: the relative position and velocity of the chaser satellite are not accurately known, the gyroscope and star tracker measurements of the chaser satellite attitude are noisy, the line-of-sight (LOS) angle measurements from the focal plane of the visible sensor are noisy, sensors may be mutually misaligned, and a reaction wheel attitude controller may be turning the chaser satellite about its roll axis, parallel to the visible sensor boresight and the laser range finder (LRF) beam, so that two solar arrays of the spacecraft, each with only one articulation degree-of-freedom about the pitch axis, maintain Sun-pointing. Furthermore, due to the occasional poor lighting of the target satellite, the visible sensor measurements can be interrupted. If this occurs, the laser range finger (LRF) pointing will be lost. But, when the visible sensor measurements resume, the required pointing accuracy must be quickly regained.

[0007] An improved technology is needed to address the above problems for relative navigation, attitude determination, pointing and tracking for spacecraft rendezvous in a circular orbit.

[0008] A closed-loop control system for autonomous spacecraft navigation is disclosed. The system and its component technologies provide improved functionality and precision for attitude determination, relative navigation, pointing, and tracking for spacecraft rendezvous in a circular orbit, or, more precisely, near circular orbit. Kalman filters of embodiments of the present invention are provided that may be applied to an architecture of one embodiment of a closed-loop system of the present invention to allow for autonomous navigation of a spacecraft even if different sensors are misaligned and measurement information is suspended, such as when a visible sensor operation is interrupted due to poor lighting of the target satellite. Kalman filters of the present invention account for, or are tolerant of, various misalignments between sensors, such as gyros, star trackers, visible sensors, and laser range finders. And because of the closed-loop integrated architecture design of one embodiment of the present invention and the versatility of the Kalman filters of the present invention, the navigation process continues even in a temporary absence of complete measurement information.

[0009] One embodiment of the present invention provides a combination of component technologies for autonomous attitude determination, relative navigation, pointing, and tracking for spacecraft rendezvous in a circular orbit. An embodiment of this combination is provided that is used to point a laser ranger finder at a target satellite with effective pointing accuracy.

[0010] One embodiment of the present invention provides an analytical technique and algorithm for correcting gyro errors using star tracker measurements for precise attitude determination. One embodiment of the present invention uses a space integrated global positioning system and inertial positioning system (Space Integrated GPS/INS or SIGI) of a chaser satellite supplemented by a steady-state three-axis correction algorithm for gyro-measured attitude determination errors using star tracker measurements.

[0011] One embodiment of the present invention provides Kalman filter relative navigation component technology. One embodiment of the present invention determines the relative position and velocity of a target satellite using a Kalman filter that accepts line-of-sight (LOS) azimuth and elevation angle measurements from visible and infrared sensor focal planes, laser range finder range measurements, if available, and attitude estimates from a six-state attitude Kalman filter.

[0012] From another embodiment of the present invention, it is shown that the in-track distance of a target satellite is observable from the elevation angle measurement of the focal plane, provided the relative altitude between the chaser satellite and the target satellite is not zero, or equivalently, the elevation angle is nonzero. Thus, a laser range finder is not a compelling requirement for long-range (˜10 km) measurements of a target satellite in one embodiment of the present invention. In yet another embodiment of the present invention, it is shown that if a laser range finder must be used nonetheless for long-range measurements of a target satellite, a collimated visible sensor can be pointed with effective accuracy using angle measurements of the visible sensor, a six-state extended Kalman filter for relative navigation, pointing commands, an attitude controller, SIGI gyros and a star tracker for attitude measurements, and a six-state Kalman filter for gyro measurement updates using star tracker measurements.

[0013] One embodiment of the present invention provides a technique to achieve virtual observability of a target in presence of visual measurement interruptions. In one embodiment of the present invention the observability of a visible sensor for estimating the range of a target from azimuth and elevation angles at long range is maintained by propagating the relative position and velocity estimation of the spacecraft just prior to the interruption, provided estimates of relative incremental velocity continue to be available from the accelerometer and simulated environment.

[0014] One embodiment of the present invention provides a technique to use the roll angle freedom of pointing commands to orient the solar arrays of a chaser satellite towards the Sun. In one embodiment of the present invention, these pointing commands refer to quaternion attitude pointing commands and angular rate commands for a chaser satellite attitude controller that may be used for any of acquiring and tracking a target satellite, pointing a laser range finder at a target satellite, and orienting a solar array towards the Sun.

[0049] The present inventions now will be described more fully hereinafter with reference to the accompanying drawings, in which some, but not all embodiments of the invention are shown. Indeed, these inventions may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Like numbers refer to like elements throughout.

[0050] While a primary use of the present invention is in the field of spacecraft rendezvous, it will be appreciated from the following description that the invention is also useful for autonomous formation flights and autonomous spacecraft docking. The terms spacecraft, satellite, and space vehicle are used interchangeably throughout this application to refer to a body in space, including but not limited to satellites, shuttles, space stations, rockets, and other orbiting spacecraft. The term circular orbit is also meant to include a near circular orbit.

[0051] The following references are cited herein and are incorporated by reference:

[0052] Ref. 1, Hablani, H. B., Tapper, M. L., and Dana-Bashian, D. J., Guidance and Relative Navigation for Autonomous Rendezvous in a Circular Orbit, 25 AIAA J. Guidance, Control, and Dynamics 3, pp. 553–62 (2000);

[0053] Ref. 2, Markely, F. L., and Reynolds, R. G., Analytic Steady-State Accuracy of a Spacecraft Attitude Estimator, 23 AIAA J. Guidance, Control, and Dynamics 6, pp. 1065–67 (2000);

[0054] Ref. 3, Wu, Y.-W., Li, R., and Robertson, A. D., Precision Attitude Determination for GOES N Satellites, AAS 03-002;

[0055] Ref. 4, Bryson, A. E., Jr., Control of Spacecraft and Aircraft, ch. 1 (Princeton University Press 1994);

[0056] Ref. 5, Reynolds, R. G., Quaternion Parameterization and a Simple Algorithm for Global Attitude Estimation, 21 AIAA J. Guidance, Control, and Dynamics 4, pp. 669–71 (1998);

[0057] Ref. 6, Markley, F. L., Fast Quaternion Attitude Estimation form Two Vector Measurements, 25 AIAA J. Guidance, Control, and Dynamics 2, pp. 411–14 (2002);

[0058] Ref. 7, Shneydor, N. A., Missile Guidance and Pursuit Kinematics, Dynamics and Control, Appendix A, pp. 217–18 (Horwood Publishing 1998);

[0059] Ref. 8, Chen, C. T., Introduction to Linear System Theory, Holt, Rinehart and Winston, Inc., New York, 1970, Sec. 5.4;

[0060] Ref. 9, Bar-Shalom, Y., Rong Li, X., and Kirubarajan, T., Estimation With Applications to Tracking and Navigation, John Wiley and Sons, New York, 2001, Sec. 3.7;

[0061] Ref. 10, Fogel, E., and Gavish, M., “n-th Dynamics Target Observability From Angle Measurements”, 24 IEEE Trans. on Aerospace and Electronic Systems, pp. 305–308 (May 1988);

[0062] Ref. 11, Kailath, T., Linear Systems, ch. 2 (Prentice-Hall 1996);

[0063] Ref. 12, Mullins, L. D., Initial Value and Two Point Boundary Value Solutions to the Clohessy-Wiltshire Equations, 40 J. Astronautical Sciences 4, pp. 487–501 (October–December 1992); and

[0064] Ref. 13, Bialke, W., High Fidelity Mathematical Modeling of Reaction Wheel Performance, AAS 98-063, pp. 483–96.

[0065] Range measurement may be effected with a laser range finder (LRF). However, because of the narrow beam of an LRF and according to current estimates, a laser range finder typically is pointed at a target satellite at a distance of 10 km with the precision of 90 μrad per axis, 1σ. The target may be first acquired with a narrow field-of-view visible camera. The visible camera and the laser range finder may be collimated. Although there may be slight misalignment between the two boresights, this misalignment generally is calibrated within an accuracy of 20 μrad (about one-fourth of the pointing accuracy requirement). The visible camera may provide the azimuth and elevation angle measurements that are used to point the camera and the laser range finder at the target within the pointing accuracy required for the laser range finder. The LRF may then measure the range of the target periodically as desired, depending on the magnitude of the range estimation error variance.

[0066] The visible camera may continue to track the target satellite in a closed-loop system of one example embodiment of the present invention using the quaternion and angular rate commands described more fully herein. These commands require a relative position vector r_CT from chaser to target and a relative velocity vector {dot over (r)}_CT. Because these vectors are not exactly known, they may be estimated using a relative six-state navigation Kalman filter described more fully herein. If the target satellite is in a circular orbit and is passive (that is, the target neither helps nor hinders an attempt to rendezvous with the target) and if the chaser satellite is sufficiently close to the target satellite such that the arc separation between the two vehicles is less than the angle measurement noise of the visible camera, then the relative Kalman filter may be based on the linear, time-invariant Clohessy-Wiltshire equations. One such filter is described in Ref. 1. If the relative distance is too large to effectively use the linear, time-invariant Clohessy-Wiltshire equations, the relative navigation may be based on a relative inertial motion formulation in an inertial frame.

[0067] One example embodiment of the present invention of the architecture required for pointing a visible camera with precision and tracking the target is shown in FIG. 1. The attitude of the chaser spacecraft, quaternion q in FIG. 1, is determined by using gyros and a star tracker (q_gyro/ST). A periodic correction of gyro-measured attitude is performed with star tracker measurements and a six-state Kalman filter for attitude determination. The delta-velocity (ΔV) firings to bring the chaser satellite closer to the target satellite, according to guidance algorithms described in Ref. 1 or other guidance algorithms, are measured by the accelerometer and inputted to a different six-state Kalman filter for estimated relative position ({circumflex over (r)}) and velocity ({dot over ({circumflex over (r)}) propagation. The gyros and accelerometers may be part of the SIGI (Space Integrated GPS/INS) navigation system for the chaser satellite. The covariance matrices of the inertial position, velocity, attitude and rate errors are considered in the six-state relative navigation Kalman filter. The attitude and rate (q, ω) of the chaser spacecraft track the quaternion commands and rates (q_C, ω_C), respectively. A proportional-integral-and-derivative (PID) attitude controller, with gains K_P, K_I, and K_D, and reaction wheels as actuators are used to track the quaternion commands and rate commands.

[0068] As shown in FIG. 1, different components of a closed-loop control system may operate at different frequencies. The frequency for a component may be any frequency that provides effective communication with the interlinked components; the frequencies presented in FIG. 1 and throughout this application are meant only as examples and are not intended as limiting the underlying present invention. Also, in the exemplary embodiments of the present invention and associated equations, the chaser is usually treated as a finite body with attitude, and the target satellite is treated as a point mass.

[0069] Another embodiment of the present invention is presented in FIG. 2a as a closed-loop concatenation of component technologies. A more detailed embodiment of the present invention is presented in FIG. 2b. Inertial navigation of the chaser satellite, and feeding the estimates of satellite position, velocity, attitude and rates and their variances to the relative navigation component in the inertial or relative frame of the target satellite are shown in greater detail in FIG. 2b. The examples of LRF pointing technology presented subsequently do not utilize the inertial relative navigation shown in FIG. 2b; instead, they are based on relative navigation in local-vertical-local-horizontal frame and Clohessey-Wiltshire equations. This illustrates that a specific component technology developed and illustrated herein as a part of the laser range finder pointing technology is replaceable by a different component technology if it serves the same purpose, albeit possibly with different accuracies. The following are descriptions of various component technologies of embodiments of the present invention.

[0070] A six-state, steady-state, analytical Kalman filter of one embodiment of the present invention estimates three-axis attitude of a chaser satellite with respect to an inertial or local-vertical-local-horizontal (LVLH) frame using gyros and a star tracker. A specific embodiment may estimate attitude of a line-of-sight or visible sensor of a chaser satellite Therefore, this component technology is presented first. The present invention provides an analytical technique and algorithm for correcting gyro errors using star tracker measurements for three-axis attitude determination. Space Integrated GPS/INS (SIGI) gyros and star trackers may determine the focal plane attitude in the inertial frame. Gyro measurements are corrupted with noise, drift and random walk error, while star tracker measurements are corrupted with noise only.

[0071] The noise, drift and random walk in the SIGI gyro measurements are corrected periodically with the star tracker measurements. For this purpose, a single-axis analysis of a Kalman filter for gyro error corrections, as presented in Ref. 2, is modified for a three-axis application. Though this pairing of hardware is conventional, the Kalman filtering of the present invention and described herein is unique in that the gyro-measured spacecraft attitude is blended with a time-matched star tracker measured attitude, and three-axis attitude estimation errors are calculated analytically before and after star tracker updates. The small three-axis attitude estimation error is separated from the large three-axis gyro and star tracker measurements and then the analytic, steady-state, attitude determination Kalman filter optimally updates the large three-axis gyro measurements and suppresses the inherent random walk and drift in the measurements. This analytically solved attitude Kalman filter facilitates a selection of the optimum star tracker measurement interval and star tracker orientation with respect to the spacecraft axes, as provided further with respect to, for example, star tracker measurement interval T, parameters S_e, S_v, and S_u, ζ, and σ_st, Eqs. (16a), (16b), (16c), (17a), and FIGS. 14a and 14b. One example embodiment of an analytical attitude determination technology of the present invention is provided below.

[0072] FIG. 3 illustrates a process of measuring spacecraft attitude with a gyro at an interval of τ at t=kτ (k=1,2, . . . , n) and with a star tracker at an interval of T=nτ. This process is provided in Ref. 2. The index k is reinitialized to 0 when it is equal to n. The three-axis incremental angle Δφ_k measured by gyros during the interval t_k−1≦t≦t_k is
Δφ_k=Δθ_k+τb_k+β_k+q_k  (1)
where Δθ_k is the true change in the spacecraft attitude, equal to

[0073] $\begin{array}{cc}\Delta {\underset{\_}{\theta }}_k={\int }_{t_{k-1}}^{t_k}\underset{\_}{\omega }dt=\tau {\underset{\_}{\omega }}_{k-1}& \left(2\right)\end{array}$
The gyro sample period τ is sufficiently small so that the above integration is correct. In the subsequent equation, ω represents the spacecraft angular rate, b is the drift rate of the gyro, β_k is a zero-mean noise in the measurement arising from a 3×1 random walk rate n_v (t) and a 3×1 drift acceleration n_u so that β_k is

[0074] $\begin{array}{cc}{\underset{\_}{\beta }}_k={\int }_{\left(k-1\right)\tau }^{k\tau }{\underset{\_}{n}}_v\left(t\right)dt+{\int }_{\left(k-1\right)\tau }^{k\tau }\left(k\tau -t\right){\underset{\_}{n}}_u\left(t\right)d\tau & \left(3\right)\end{array}$
Its variance is a 3×3 diagonal matrix in which each element will appear as
σ_β²=σ_v²τ+σ_u²τ³/3  (4)
where σ_v²(rad²/s) and σ_u²(rad²/s³) are power spectral densities of the scalar elements of n_v and n_u. See Ref. 2. The 3×1 vector q_k in Eq. (1) is a non-additive quantization noise in the measurement. It is non-accumulative in that each successive addition of the angle Δφ_k does not cause this noise to accumulate, as does the noise β_k. Although a quantization noise is not a white noise, it is treated so here and regarded as a readout noise of power spectral density equal to σ_e²=q²/12 where q denotes the angle per quantum signal of the gyro.

[0075] In the duration, t_k−1≦t≦t_k, the drift rate b may also change, and this change is modeled, in vector notations, as
b_k=b_k−1+α_k α_k=N(0, σ_α)  (5)
where α_k is a zero mean, discrete random rate noise vector equal to

[0076] $\begin{array}{cc}{\underset{\_}{\alpha }}_k={\int }_{\left(k-1\right)\tau }^{k\tau }{\underset{\_}{n}}_u\left(t\right)dt& \left(6\right)\end{array}$
and the variance of each of its element is σ_α²=σ_u²τ. See Ref. 2.

[0077] The estimate of the incremental angle Δθ_k, denoted Δ{circumflex over (θ)}_k, is obtained from the gyro output Δφ_k; thus
Δ{circumflex over (θ)}_k=Δφ_k−τ{circumflex over (b)}_k  (7a)
{circumflex over (b)}_k={circumflex over (b)}_k−1  (7b)
where {circumflex over (b)}_k is the estimate of the drift rate b_k, estimated with the aid of star tracker measurements. See Ref. 2.

[0078] Inertial attitude of the spacecraft is obtained recursively as follows
Ĉ_kI=Ĉ_k,k−1Ĉ_k−1,I  (8)
where, in terms of Δ{circumflex over (θ)}_k and its antisymmetric matrix counterpart Δ{circumflex over (θ)}_k^x, the incremental direction cosine matrix is

[0079] $\begin{array}{cc}{\widehat{\underset{\_}{C}}}_{k,k-1}=1-\Delta {\widehat{\underset{\_}{\theta }}}_k^x+\left[\Delta {\widehat{\underset{\_}{\theta }}}_k\Delta {\widehat{\underset{\_}{\theta }}}_k^T-{\Delta {\widehat{\underset{\_}{\theta }}}_k}^2\underset{\_}{1}\right]/2& \left(9\right)\end{array}$
and 1 is a 3×3 identity matrix. In Eq. (8), Ĉ_kI is the estimate of the spacecraft attitude C_kI at t=t_k relative to an inertial frame. When k=n, one or more star trackers provide spacecraft attitude equal to C_n.st,I relative to the inertial frame. This involves a 3×1 star tracker angle noise vector, ν_st; thus
C_n.st,I=(1−ν_st^x)C_nI  (10)
where C_nI is the true spacecraft attitude at k=n, and t=t_n. The three elements of the zero-mean random Gaussian noise vector ν_st may all be different, depending on the number of star trackers and their orientations relative to the spacecraft axes. The low- and high-frequency spatial noise in star tracker measurements (Ref. 3) may be ignored, as it is in this embodiment of the present invention.

[0080] Because of the noise, drift and random walk errors, the gyro measurements are updated using star tracker measurements.

[0081] Ref. 2 shows that, in single-axis, the steady-state correction of the gyro-measured attitude with star tracker measurements at k=n may be performed as follows. Let {circumflex over (θ)}_nΔ{circumflex over (θ)}₀(−) be the gyro estimated spacecraft attitude just prior to the star tracker measurement, and let θ_st be the spacecraft attitude measured by the star tracker at that instant. Then the estimate {circumflex over (θ)}₀(−) is corrected to {circumflex over (θ)}₀(+) such that
{circumflex over (θ)}₀(+)={circumflex over (θ)}₀(−)+(1−ζ⁻²)[θ_st−{circumflex over (θ)}₀(−)]  (11)
where the parameter ζ is defined below. Recall that the subscript n is replaced with 0 and the star tracker update is performed periodically at k=n with the period T=nτ.

[0082] A three-axis equivalent of Eq. (11) is developed as follows. Eq. (8) provides the gyro estimated spacecraft attitude Ĉ_n.gyro,I at k=n, where the subscript gyro is added to distinguish this attitude matrix from the star tracker measured attitude matrix C_n.st,I given by Eq. (10). Similar to Eq. (10), the gyro estimated inertial attitude has an accumulated estimation error equal to ν_n.gyro, and, therefore,
Ĉ_n.gyro,I=(1−{circumflex over (ν)}_n.gyro^x)C_nI  (12)
Since the true attitude C_nI is unknown, the estimation error angle {circumflex over (ν)}_n.gyro is also unknown.

[0083] The three axis equivalent of the quantity θ_st−{circumflex over (θ)}₀(−) in Eq. (11) is
θ_st−{circumflex over (θ)}₀(−)C_n.st,IĈ_I,n.gyro  (13a)

[0084] $\begin{array}{cc}\approx \underset{\_}{1}-{\left({\underset{\_}{v}}_{\mathrm{st}}-{\underset{\_}{\hat{v}}}_{n.\mathrm{gyro}}\right)}^x& \left(13b\right)\\ \stackrel{\Delta }{=}\underset{\_}{1}-{\underset{\_}{\hat{v}}}_{\mathrm{st}/\mathrm{gyro}}^x& \left(13c\right)\end{array}$
Since both C_n.st.I and Ĉ_n.gyro,I are known, the multiplication in (13a) may be performed and the small angle error vector {circumflex over (ν)}_st/gyro defined by Eqs. (13b) and (13c) may be calculated. The error angle {circumflex over (ν)}_st/gyro is a 3×1 relative error angle vector between the star tracker measurements and the gyro-estimated attitude.

[0085] To incorporate the star tracker update factor (1−ζ⁻²) of Eq. (11) in the three-axis attitude estimation technique, it is recognized that the parameter ζ, defined later, depends on the star tracker measurement noise for the axis at hand. Hence, each of the three elements ν_st/gyro,α (α=x,y,z) of ν_st/gyro is corrected to (1−ζ_α⁻²)ν_st/gyro,α. To abbreviate notations, introduce

[0086] ${\underset{\_}{v}}_{\mathrm{st}/\mathrm{update}}\stackrel{\Delta }{=}\left[\begin{array}{cc}\left(1-{\zeta }_x^{-2}\right)& v_{\mathrm{st}/\mathrm{gyro},x}\\ \left(1-{\zeta }_y^{-2}\right)& v_{\mathrm{st}/\mathrm{gyro},y}\\ \left(1-{\zeta }_z^{-2}\right)& v_{\mathrm{st}/\mathrm{gyro},z}\end{array}\right]$
Then, the following three-axis equivalent of Eq. (11) can be written by inspection.
Ĉ_0.gyro,I(+)=[1−ν_st/update^x]Ĉ_n.gyro,I  (14)
A single-axis equivalent of the above correction process is graphically presented in FIG. 4.

[0087] With regard to the gyro bias and using a single-axis update model of Ref. 2, the three-axis update of the bias vector

[0088] ${\underset{\_}{\hat{b}}}_n\stackrel{\Delta }{=}{\underset{\_}{\hat{b}}}_0(-)$
is

[0089] $\begin{array}{cc}{\underset{\_}{\hat{b}}}_0(+)={\underset{\_}{\hat{b}}}_0(-)-\left[\begin{array}{cc}{S}_{u,x}\left({\zeta }_{x}T\right)-1& v_{\mathrm{st}/\mathrm{gyro},x}\\ S_{u,y}\left({\zeta }_{y}T\right)-1& v_{\mathrm{st}/\mathrm{gyro},y}\\ S_{u,z}\left({\zeta }_{z}T\right)-1& v_{\mathrm{st}/\mathrm{gyro},z}\end{array}\right]& \left(15\right)\end{array}$
where the dimensionless parameter S_u (of which there are three, one for each axis) is defined below.

[0090] To compare gyro noise with the star tracker noise, Ref. 2 introduces the following three dimensionless parameters for each axis of the spacecraft
readout noise: S_e=σ_e/σ_st  (16a)
random walk noise: S_v=T^1/2σ_v/σ_st  (16b)
drift angle: S_u=T^3/2σ_u/σ_st  (16c)
The parameter ζ is defined in terms of the above three dimensionless parameters

[0091] $\begin{array}{cc}\zeta =\gamma +\frac{1}{4}{S}_u+\frac{1}{2}{\left(2{\mathrm{\gamma S}}_u+S_v^2+\frac{1}{3}{S}_u^2\right)}^{1/2}\mathrm{where}& \left(17a\right)\\ \gamma ={\left(1+S_e^2+\frac{1}{4}{S}_v^2+\frac{1}{48}{S}_u^2\right)}^{1/2}\ge 1& \left(17b\right)\end{array}$
In the above algorithm, a six-state Kalman filter is literally solved. The covariances of the attitude and bias estimation errors and their correlation are incorporated herein from Ref. 2.

[0092] A relative navigation extended Kalman filter (EKF) of one embodiment of the present invention provides the position and velocity estimates of a chaser satellite relative to a target satellite required for the calculation of pointing and rate commands for tracking the target and guidance delta-velocities for rendezvous and proximity operations.

[0093] The optimum three-axis spacecraft local-vertical local-horizontal (LVLH) attitude estimate, determined from the previous section, is used recursively in a six-state relative navigation extended Kalman filter (EKF) of one embodiment of the present invention. This relative navigation Kalman filter uses the line-of-sight (LOS) azimuth and elevation angle measurements from the visible sensor focal plane.

[0094] The measurements for a relative navigation filter of one embodiment of the present invention come from two sources: the Space Integrated GPS/INS (SIGI) gyros and accelerometers for inertial navigation of the chaser satellite and a rendezvous sensor suite including visible and infrared sensors and a laser range finder. The visible and infrared sensors provide angular location of the target relative to their focal plane frame. To achieve the tight pointing accuracy, this relative navigation filter takes into account the visible sensor focal plane offset from the chaser satellite center of mass and the known misalignment between the visible sensor and the LRF.

[0095] A relationship is established between partial derivatives of the focal plane angle measurements and relative position of the target explicitly through the three-axis gyro/star tracker estimated attitude transformation matrix. One embodiment of focal plane and range measurements of the present invention are provided below.

[0096] FIG. 5a illustrates relative geometry of chaser and target satellites, a focal plane of an imaging sensor of the chaser satellite, and an image of a target satellite on the focal plane. Because of tracking errors, the target image may not be at the center of the focal plane. Instead, it may be at nonzero azimuth and elevation angles with respect to the center of the focal plane. Let the line-of-sight (LOS) vector l, expressed in the misaligned sensor frame F′_S, have the components l^F′S=[x′_Sy′_Sz′_S]^T. The line-of-sight (LOS) angle measurements, then, are related to the components of l by
azimuth α_m=α+ν_α=tan⁻¹(y′_S,x′_S)+ν_α  (18a)
elevation ε_m=ε+ν_ε=tan⁻¹(−z′_S, x′_S)+ν_ε  (18b)
These angles are small, much less than 1 rad, so that the target image remains within the field-of-view of the sensor. The quantities ν_α and ν_ε are zero mean, random white noise in the measurements. Using these measurements, the chaser spacecraft may be accurately pointed to the target so that a collimated laser beam from a laser range finder (LRF) can impinge upon the target satellite and determine the relative range of the target. This range measurement is modeled as
r_m=∥l∥+ν_r  (19)
where ν_r is a zero-mean range measurement white noise.

[0097] Using these angle measurements and, when available, a range measurement, a Kalman filter may estimate position and velocity of the target satellite relative to the chaser satellite. The orbital motion of the two satellites that may be expressed in Cartesian frames governs the relative position and velocity. At short distances between the two satellites, the relative motion may be modeled by Clohessey-Wiltshire (CW) equations which involve the relative position vector r_CW from the target satellite mass center to the chaser satellite mass center and its time derivatives. See Ref. 4. Clohessey-Wiltshire equations are valid for a target satellite in a circular orbit and a chaser satellite in proximity (˜10 km) such that the orbital arc separation between the two satellites is less than the standard deviation of the angle measurement noise ν_α and ν_ε introduced above. The relative position vector r_CW is expressed in the target LVLH orbit frame F₀, and, therefore, r_CW is related to the previously introduced position vectors and transformation matrices as follows

[0098] $\begin{array}{cc}{\underset{\_}{r}}_{\mathrm{CW}}^{F^0}=-{\underset{\_}{C}}_{01}{\underset{\_}{C}}_{1b}\left({\underset{\_}{C}}_{\mathrm{bs}}{\underset{\_}{C}}_{{\mathrm{ss}}^{\prime }}{\underset{\_}{l}}^{F_s^{\prime }}+{\underset{\_}{r}}_{\mathrm{off}}^{F^b}\right)=-{\underset{\_}{C}}_{0I}{\underset{\_}{r}}_{\mathrm{CT}}^{F^I}& \left(20\right)\end{array}$
where C_1b is the transformation matrix from the chaser body frame to the chaser LVLH frame F₁. C₀₁ transforms a vector in the frame F₁ to the target LVLH frame F₀ and C_0I transforms a vector in an inertial frame to the frame F₀. The negative signs in the right sides of Eq. (20) arise because the vector r_CW is from the target satellite to the chaser satellite, not from the chaser satellite to the target satellite. These matrices will be determined by the chaser satellite navigation and attitude determination system (as shown in the previous section) and initial ground tracking of the target satellite. Thus, navigation and attitude determination errors will affect the implementation of the earlier equations.

[0099] Because the line-of-sight (LOS) vector l^F′S in Eq. (20) equals

[0100] ${\left[x_s^{\prime }{y}_s^{\prime }{z}_s^{\prime }\right]}^T,$
the vector

[0101] ${\underset{\_}{r}}_{\mathrm{CW}}^{F_0}$
is related to the measurements α_m, ε_m, and r_m. Specifically, since this embodiment of the present invention uses an extended Kalman filter (EKF), the following partial derivatives of the arctan functions in the measurement Eqs. (18) and (19) are required

[0102] $\begin{array}{cc}\begin{array}{c}\frac{\partial \alpha }{\partial \underset{\_}{l}}={\left[\frac{\partial \alpha }{\partial x_s^{\prime }}\frac{\partial \alpha }{\partial y_s^{\prime }}\frac{\partial \alpha }{\partial z_s^{\prime }}\right]}^T={\left[-y_s^{\prime }{x}_s^{\prime }0\right]}^T/\left({x^{\prime }}_s^2+y_s^{\mathrm{\prime 2}}\right)\\ \frac{\partial \varepsilon }{\partial \underset{\_}{l}}={\left[\frac{\partial \varepsilon }{\partial x_s^{\prime }}\frac{\partial \varepsilon }{\partial y_s^{\prime }}\frac{\partial \varepsilon }{\partial z_s^{\prime }}\right]}^T={\left[z_s^{\prime }0-x_s^{\prime }\right]}^T/\left({x^{\prime }}_s^2+z_s^{\mathrm{\prime 2}}\right)\\ \frac{\partial l}{\partial \underset{\_}{l}}={\left[\frac{\partial l}{\partial x_s^{\prime }}\frac{\partial l}{\partial y_s^{\prime }}\frac{\partial l}{\partial z_s^{\prime }}\right]}^T={\left[x_s^{\prime }{y}_s^{\prime }{z}_s^{\prime }\right]}^T/l\end{array}& \left(21\right)\end{array}$
where the LOS range l is equal to ∥l∥ and the superscript T denotes transpose of the vector. In Eq. (21), the derivative of a scalar with respect to a vector is expressed in the form of a column vector instead of a row vector. The extended Kalman filter requires partial derivatives of α, ε and l with respect to the vector

[0103] ${\underset{\_}{r}}_{\mathrm{CW}}^{F_0}.$
Considering Eq. (20), we obtain

[0104] $\begin{array}{cc}\frac{\partial \alpha }{\partial {\underset{\_}{r}}_{\mathrm{CW}}}=-C_{0s^{\prime }}\frac{\partial \alpha }{\partial \underset{\_}{l}}& \left(22a\right)\\ \frac{\partial \varepsilon }{\partial {\underset{\_}{r}}_{\mathrm{CW}}}=-C_{0s^{\prime }}\frac{\partial \varepsilon }{\partial \underset{\_}{l}}& \left(22b\right)\\ \frac{\partial l}{\partial {\underset{\_}{r}}_{\mathrm{CW}}}=-C_{0s^{\prime }}\frac{\partial l}{\partial \underset{\_}{l}}& \left(22c\right)\end{array}$
where C_0s′=C₀₁C_1bC_bsC_ss′. Defining the state vector of the extended Kalman filter as

[0105] ${\left[{\underset{\_}{r}}_{\mathrm{CW}}^T,{\stackrel{.}{\underset{\_}{r}}}_{\mathrm{CW}}^T\right]}^T$
and the measurement vector

[0106] $\underset{\_}{z}={\left[{\alpha }_m{\varepsilon }_m{r}_m\right]}^T,$
the corresponding linearized (3×6) measurement matrix H is

[0107] $\begin{array}{cc}\underset{\_}{H}=\left[\begin{array}{cc}{\left(\frac{\partial \alpha }{\partial {\underset{\_}{r}}_{\mathrm{CW}}}\right)}^T& \\ {\left(\frac{\partial \varepsilon }{\partial {\underset{\_}{r}}_{\mathrm{CW}}}\right)}^T& 0_{3\times 3}\\ {\left(\frac{\partial l}{\partial {\underset{\_}{r}}_{\mathrm{CW}}}\right)}^T& \end{array}\right]& \left(23\right)\end{array}$
where 0_3×3 is a 3×3 null matrix. This H matrix is used in the extended Kalman filter. Because the matrices C₀₁ and C_1b are determined by a navigation system including gyros, accelerometers, and a GPS receiver, the performance of the Kalman filter is influenced by navigation errors.

[0108] A succinct three-axis attitude and rate command component of one embodiment of the present invention is disclosed, based on Refs. 5, 6 and 7, that (1) provides quaternion pointing commands to acquire and track a target satellite and command a visible sensor boresight, initially oriented in any direction in an inertial frame, to point at the target satellite; and (2) provides angular rate commands to track the target satellite with a tighter pointing accuracy than without rate commands.

[0109] The relative position of a target estimated for relative navigation establishes the commanded line-of-sight (LOS) vector orientation of the visible sensor boresight. Commanded line-of-sight (LOS) vector orientations lead to quaternion pointing commands. See Refs. 5, 6. The disclosed quaternion commands of one embodiment of the present invention include an arbitrary angle of rotation about the line-of-sight (LOS) vector for Sun-pointing of solar arrays, and account for the offset of the sensors from the center of mass of the chaser satellite and misalignment of the sensor frames relative to the chaser satellite frame. The angular rate command for tracking a target satellite of one embodiment of the present invention uses the estimates of both relative position and velocity from the Relative Navigation Kalman filter of one embodiment of the present invention, and its formulation is based on Ref. 7.

[0110] Using these pointing commands and associated rate commands in an embodiment of the present invention of a concatenation of component navigation technologies that account for navigation and attitude determination errors, a proportional-integral-derivative (PID) attitude controller may point the visible sensor at a target spacecraft in a circular orbit with a pointing accuracy of 90 μrad or more tightly. This accuracy may be required to impinge upon a target with a narrow beam of a collimated laser range finder (or misaligned by an angle known within some precision, possibly 20 μrad) to measure the relative range to the target.

[0111] Example embodiments for deriving quaternion pointing commands and angular rate commands are provided below.

[0112] FIG. 5a shows a target satellite in a circular orbit. The local-vertical local-horizontal (LVLH) frame at the target satellite location is denoted F₀. FIG. 5a also shows a chaser satellite in a different orbit, and the local-vertical local-horizontal (LVLH) frame at the location of the chaser satellite is denoted F₁. Quaternion pointing commands of one embodiment of the present invention are attitude commands for a chaser satellite to orient the boresight of an optical or infrared sensor at a target satellite. The optical sensor may be misaligned from its ideal orientation relative to the chaser satellite body frame. The misalignment angles may be determined through calibration. Let F_S and F′_S be the ideal and the misaligned frames of the sensor, respectively, and let C_SS′ transform a vector in F′_S to the frame F_S. In the misaligned frame F′_S: x′_Sy′_Sz′_S, the boresight of the sensor is along the x′_S axis. The objective is to point the visible sensor boresight at the target satellite. Let b be a unit vector along the boresight axis. Then, in the frame F′_S: b^F′S[1 0 0]. This unit vector is transformed to the chaser body frame F_b via the transformation matrix C_SS′ and the matrix C_bs between the ideal sensor frame F_S and the chaser satellite frame F_b. Thus the boresight unit vector b in the frame F_b is
b^Fb=C_bsC_SS′b^F′S   (24)
The inertial position of the center of mass of the target satellite is denoted r₀, and the inertial position of the center of mass of the chaser satellite is denoted r₁. The relative vector r_CT from the chaser center of mass to the target center of mass, expressed in an inertial frame, is given by

[0113] $\begin{array}{cc}{\underset{\_}{r}}_{\mathrm{CT}}^{F_I}=-{\underset{\_}{r}}_1^{F_I}+{\underset{\_}{r}}_0^{F_I}& \left(25\right)\end{array}$
where r₁ and r₀ are expressed in the inertial frame F_I. In the embodiment of the present invention shown in FIG. 5a, the optical sensor is located at an offset vector, r_off, from the chaser center of mass. Thus, the line-of-sight vector from the center of the focal plane of the chaser to the target is given by l=r_CT−C_Ibr_off, where C_Ib is the transformation matrix from the chaser body frame F_b to the inertial frame F_I. A unit vector u along this line-of-sight vector is given by
u^FI=l/∥l∥  (26)

[0114] Assuming that the boresight unit vector b^Fb is known in the chaser body frame and the line-of-sight (LOS) unit vector u^FI is known in the inertial frame, the quaternion command q_cI that transforms the unit vector u^FI into the boresight unit vector b^Fb is given by

[0115] $\begin{array}{cc}{q}_{\mathrm{c1}}=\frac{1}{\sqrt{2\left(1+\underset{\_}{b}·\underset{\_}{u}\right)}}\left[\left(\underset{\_}{b}\times \underset{\_}{u}\right)\cos \frac{\varphi }{2}+\left(\underset{\_}{b}+\underset{\_}{u}\right)\sin \frac{\varphi }{2},\left(1+\underset{\_}{b}·\underset{\_}{u}\right)\cos \frac{\varphi }{2}\right]& \left(27\right)\end{array}$
where the subscript cI implies a transformation from the inertial to the commanded chaser body frame, and φ is a roll angle about the boresight or line-of-sight (LOS) vector for pointing the solar arrays to the Sun. See Refs. 5, 6. Even though the vectors b and u in Eqs. (24) and (26) are in different frames, the vector operation in Eq. (27) occurs as though b and u were in the same frame, implying that the spacecraft body frame F_b is initially aligned with the inertial frame and the command quaternion q_cI will rotate the sensor boresight to point at the target.

[0116] The angle φ for the roll angle about the boresight or line-of-sight (LOS) vector may be utilized to orient the solar arrays normal to the Sun. The degree of freedom in the quaternion pointing commands of angle φ allow the architecture of a navigation system of the present invention to simultaneously point a visible sensor towards a target satellite while maintaining solar arrays of the chaser satellite normal to the Sun. Sun-pointing commands may include both pitch and roll commands for the solar array controllers.

[0117] To obtain angular rate commands, the line-of-sight (LOS) vector l^FI from the center of the focal plane of the chaser to the target mass center given above is transformed to the chaser spacecraft body frame, to get

[0118] $\begin{array}{cc}{\underset{\_}{l}}^{F}b={\underset{\_}{r}}_{\mathrm{CT}}^{F_b}-{\underset{\_}{r}}_{\mathrm{off}}^{F_b}& \left(28\right)\end{array}$
The relative inertial velocity of the target with respect to the chaser satellite is

[0119] $\begin{array}{cc}{\stackrel{.}{\underset{\_}{r}}}_{\mathrm{CT}}^{F_I}={\stackrel{.}{\underset{\_}{r}}}_0^{F^I}-{\stackrel{.}{\underset{\_}{r}}}_1^{F^I}& \left(29\right)\end{array}$
where a dot (•) over a vector represents the inertial derivative of the vector. The relative inertial velocity may be estimated using a Kalman filter. In the chaser body frame, this relative inertial velocity vector will be

[0120] ${\stackrel{.}{\underset{\_}{r}}}_{\mathrm{CT}}^{F_b}={\underset{\_}{C}}_{\mathrm{bI}}{\stackrel{.}{\underset{\_}{r}}}_{\mathrm{CT}}^{F_I}.$
The inertial velocity of the body-fixed vector

[0121] ${\underset{\_}{r}}_{\mathrm{off}}^{F_b}$
in the chaser frame

[0122] $F_b\mathrm{is}{\stackrel{.}{\underset{\_}{r}}}_{\mathrm{off}}^{F_b}={\underset{\_}{\omega }}_{\mathrm{bI}}\times {\underset{\_}{r}}_{\mathrm{off}}^{F_b}$
where ω_bI is the inertial angular velocity of the chaser body frame. The inertial velocity of the line-of-sight vector l^Fb is then

[0123] $\begin{array}{cc}{\stackrel{.}{\underset{\_}{l}}}^{F_b}={\underset{\_}{C}}_{\mathrm{bI}}{\underset{\_}{\stackrel{.}{r}}}_{\mathrm{CT}}^{F_I}-{\underset{\_}{\omega }}_{\mathrm{bI}}\times {\underset{\_}{r}}_{\mathrm{off}}^{F_b}& \left(30\right)\end{array}$
Knowing {dot over (l)}^Fb, the inertial angular velocity of the line-of-sight (LOS) vector l^Fb is given by
ω_cI=l^Fb×{dot over (l)}^Fb/(l·l)  (31)
ω_cI is the rate command for the chaser body frame expressed in the current body frame. See Ref. 7. The quaternion pointing command, Eq. (27), and the angular rate command, Eq. (31), may be used to acquire and track a target satellite, as described with respect to a closed-loop laser range finder pointing architecture as one embodiment of the present invention.

[0124] In the first part of this section, it is shown that the relative range of a target satellite is observable from the azimuth and elevation angle measurements of an embodiment of the present invention formulated in a previous section, provided that there is a nonzero relative altitude between the two satellites. In the second part of this section, the consequences of interruption of the angle measurements are determined. The angle measurements of the visible sensor may be interrupted when the target satellite is in shadow and the Sun's rays do not illuminate the target. If the visible sensor measurements are interrupted for a finite duration, the variance of the position estimation error grows. Because of the stringent pointing accuracy required for a laser range finder and the ensuing complexities, the present invention ensures that the closed-loop system will continue to perform if the observability of relative range from visible and infrared sensor focal plane angle measurements is impaired, and that the system resumes fully when the target is regained.

[0125] When the line-of-sight (LOS) angle measurements resume, the relative position vector range is rendered observable again, and the relative navigation Kalman filter quickly decreases the position (range) and velocity estimation errors to the no-interruption level as the chaser vehicle approaches the target satellite, that is the level the position and velocity errors would have been at that time if the line-of-sight measurements would not have been interrupted. Therefore, accurate range measurement from the laser range finder may not be required, resulting in a system that may cost less. Because the observability of range is through the arctan functional relationships between the coordinates [x y z] in the LVLH frame and angles, observability is not instantaneous or differential. See Ref. 8.

[0126] One embodiment of observability testing of the present invention is presented below to show that the relative position vector range is observable from angle measurements.

[0127] To examine observability of the relative range, the state space matrix associated with the Clohessey-Wiltshire (CW) equations is needed. FIG. 5b depicts a chaser vehicle in proximity of a target vehicle and illustrates the local-vertical local-horizontal (LVLH) frame F₀: x, y, z at the target mass center, where x is along the velocity vector, z points to the center of mass of the Earth, and y is normal to the orbit plane, completing the right handed triad; see also FIG. 5a. The chaser is located at (x, y, z) and its relative inertial velocity in the LVLH frame is [u v w]^T where u={dot over (x)}−ωz, v={dot over (y)}, and w=ż+ωx, and (−ω) is the orbital angular velocity of the target satellite about the y-axis. See Ref. 4. The cross-track motion y normal to the orbit plane is decoupled from the two-dimensional (x, z) motion in the orbit plane. For simplicity, this example only presents the coupled in-plane motion. The 4×1 state vector x=[u w x z]^T is governed by a first-order state-space equation. See Ref. 4. The associated 4×4 state-space matrix A is

[0128] $\begin{array}{cc}\underset{\_}{A}=\left[\begin{array}{cccc}0& \omega & -{\omega }^2& 0\\ -\omega & 0& 0& 2{\omega }^2\\ 1& 0& 0& \omega \\ 0& 1& -\omega & 0\end{array}\right]& \left(32\right)\end{array}$

[0129] A visible sensor mounted on the chaser satellite senses the target satellite, and measures the location of the centroid of the target image on the focal plane. For this example, the two vehicles are assumed to be sufficiently close together so the LVLH frame F₁ at the chaser satellite center-of-mass shown in FIG. 5a may be parallel to the frame F₀: x y z. The line-of-sight (LOS) unit vector defined by the focal plane measurements in the focal plane coordinates is transformed into the LVLH frame F₁. In the F₁ frame, the pitch angle measurement associated with the relative altitude z, then, will be given by
θ_y,m=tan⁻¹(−z/x)+ν_y   (33)
where ν_y is a zero-mean white Gaussian measurement noise. Unlike azimuth and elevation angles of Eqs. (18), the pitch angle, Eq. (33), will be large. To use this nonlinear measurement in a Kalman filter framework, the Taylor series expansion for Eq. (33) about the estimated quantities is noted:
θ_y,m≈ tan⁻¹(−{circumflex over (z)}/{circumflex over (x)})+H(x−{circumflex over (x)})+ν_y  (34)
where, in standard Kalman filter notations, the symbol ({circumflex over (.)}) denotes an estimate of the variable (.) before measurement updates and the 1×4 row matrix H is defined by
H=∂θ_y/∂x=[00 {circumflex over (z)}−{circumflex over (x)}]/ρ²  (35)
with ρ²={circumflex over (x)}²+{circumflex over (z)}².

[0130] To determine observability of the range or in-track distance x₀ between the two vehicles from the nonlinear angle measurement θ_y,m, the “maximum likelihood” estimation technique that operates on a batch of measurements may be used and the initial value of the coordinate vector [x y z]^T may be estimated. See Ref. 9, 10. However, in this embodiment, observability is established using the state space matrix A, Eq. (32), and the linearized measurement matrix H, Eq. (35). See Ref. 8.

[0131] A linear time-invariant, nth-order deterministic system is observable if the rank of its n×n observability matrix O_m involving A and H is n, or, equivalently, if the determinant O_m≠0. For A and H shown above, the observability matrix O_m is

[0132] $\begin{array}{cc}{\underset{\_}{O}}_m={\rho }^{-2}\left[\begin{array}{cccc}0& z& 2\omega x& -4{\omega }^{2}z\\ 0& -x& 2\omega z& {\omega }^{2}x\\ z& \omega x& -2{\omega }^{2}z& -{\omega }^{3}x\\ -x& \omega z& -{\omega }^{2}x& 2{\omega }^{3}z\end{array}\right]& \left(36\right)\end{array}$
and its determinant is
det O_m=3ω⁴z²(x²+4z²)/(x²+z²)  (37)
which is not zero unless the relative altitude z=0. When z=0, the chaser and the target satellites are on the same orbit, the chaser either ahead or behind the target. In these instances, the pitch angle is either zero or π rad, and it will not provide information concerning the relative distance between the chaser and the target satellite. Thus, for the range between the chaser satellite and the target satellite to be observable, the altitude z may not be zero.

[0133] The example observability test presented above does not explicitly determine the observable and the unobservable modes of natural motion of the chaser satellite in proximity of a passive target satellite. To do so, the eigenvector test of observability presented in Ref. 11 may be used, according to which, if p_i (i=1, . . . , n) is an eigenvector of the n×n state-space matrix A, then, for scalar measurements (H=1×n), the scalar Hp_i must be nonzero for observability of the vector p_i. The relative motion of the chaser satellite consists of four modes: two complex conjugate modes associated with elliptic circumnavigation, a drift mode at a constant relative altitude, and a stationary mode at a constant angular distance from the target satellite in the same orbit. See Ref. 4. The example above only shows the stationary mode to be unobservable.

[0134] The two real eigenvectors, p₁ and p₂, associated with the two complex conjugate modes that govern elliptic circumnavigation are p₁=[0 ω 2 0]^T and p₂=[ω 0 0 1]^T. See Ref. 4. The corresponding Hp₁ and Hp₂ are 2z (x²+z²)⁻¹ and −x(x²+z²)⁻¹ which are not zero simultaneously for an elliptic flyaround. Thus, the position vector during elliptic circumnavigation is observable from angle measurements. The station-keeping mode p₃ at a constant x and zero z is p₃=[0 ω 1 0]^T. The corresponding Hp₃=z/(x²+z²)=0 because z=0. Hence, as concluded earlier, the stationary mode is unobservable. The station-keeping mode and the drift mode are at a constant relative altitude both associated with a zero eigenvalue of multiplicity two. As such, the drift has the principal eigenvector p₃=[0 ω 1 0]^T and a secondary eigenvector p₄=[ω/3 0 0 ⅔]^T. While Hp₃ was just given, Hp₄=−(⅔)×(x²+z²)⁻¹. Since the drift takes place at a constant nonzero relative altitude, Hp₄ is nonzero, and, therefore, the drift mode is observable from angle measurements.

[0135] Thus, as shown by the analysis above, so long as there is a relative altitude between the chaser and target satellites, the relative position vector (and therefore relative range) is observable from the in-plane angle measurements.

[0136] Provided below are example numerical results and further description of embodiments of the present invention. Associated figures provide illustrations and results for embodiments of the present invention of guidance and navigation for three scenarios: a far-field rendezvous at approximately 200 km without a laser range finder; laser ranger finder pointing accuracy at medium range (˜10 km) without sensor misalignments; and the laser rangefinder pointing accuracy in presence of misalignment of all sensors.

[0137] In the first scenario, although the Clohessey-Wiltshire equations are not appropriate at a relative in-track distance of 200 km for satellites at 500 km altitude, the conclusions that will be drawn based on these results hold true even when the relative motion is modeled in an inertial frame or the distance between the two spacecraft is shorter (˜10 km).

[0138] The objective of the first scenario is to show observability of the range from the angle measurements. This avoids the necessity of a long-range laser range finder (LRF) and the attitude control hardware necessary for the tight pointing to a target. An additional objective of this scenario is to show that if the visible sensor measurements are interrupted due to poor lighting of the target satellite, the range estimation error does not grow in that period provided the accelerometers continue to measure acceleration of the chaser satellite and there is an estimate of any other unmeasured relative acceleration between the two spacecraft.

[0139] If a laser range finder (LRF) is used at a range of ˜10 km, a second scenario illustrates the performance of a tight attitude control system to point a visible sensor at a target satellite using angle measurements of the sensor, in presence of the sensor noise, SIGI navigation errors, and gyro/star tracker attitude determination errors. In this scenario, all sensors are assumed to be aligned perfectly.

[0140] The third scenario considers misalignments between the visible sensor, laser rangefinder, gyros and star tracker. The objective of this illustration is to determine the limits of various misalignments without violating the pointing accuracy requirements.

[0141] The three scenarios use a rendezvous sensor suite comprising a visible sensor, an infrared sensor, and a laser rangefinder. Their range of operation and noise characteristics are presented in Table 1. In each scenario, it will be stated which sensor(s) is (are) used.

[0142]

TABLE 1
Visible, IR, and Lidar Sensor Parameters
	Visible Sensor	IR Sensor	Laser Range Finder
Acquisition	≦200 km	≦20 km	≦10 km
Range
LOS Angle	0.182 mrad (1σ)	200 ≧ r ≧ 11 km	1.02 mrad (1σ)	20 ≧ r ≧ 2 km
Measurement	0.182 → 20 mrad (1σ)	11 ≧ r ≧ 0.1 km	1.02 → 20 mrad (1σ)	2 ≧ r ≧ 0.1 km
Noise
Range	0.1 r (1σ)	0.2 ≧ r ≧ 0.001 km	0.1 r (1σ)	0.1 ≧ r ≧ 0.001 km	100 → 20 m (1σ)	10 ≧ r ≧ 0.2 km
Measurement					5 m (1σ)	0.2 ≧ r ≧ 0.1 km
Noise

[0143] FIG. 6 illustrates a nominal scenario of a far-field rendezvous starting from ˜200 km behind and 8 km below a target satellite and the associated timeline. Specifically, the chaser starts at a relative position of [−200 0 8]^T km in the LVLH- frame at t=0 with a relative velocity {dot over (x)}=3ω_orb Z in the LVLH frame so that the chaser drifts at a constant altitude towards the target satellite until it reaches the coordinates [−28 0 8] km from the target where {dot over (y)}=0, ż=0. With a suitable increment ΔV at this location, the chaser is navigated to −6 km behind the target at zero relative altitude, and then to −1 km to +0.12 km with additional ΔVs, each after a transit period sufficiently different from the singular transit periods discussed in Ref. 12. The chaser attitude determination and control system are not simulated in this scenario. For initial conditions in the simulation, the above nominal scenario is perturbed. The true coordinates of the chaser satellite, r, and its estimate, {circumflex over (r)}=[{circumflex over (x)} ŷ {circumflex over (z)}]^T, are initialized to

[0144] $\begin{array}{cc}\underset{\_}{r}=\left[\begin{array}{c}-200+5\\ 0+2\\ 8+2\end{array}\right]\mathrm{km},\widehat{\underset{\_}{r}}=\left[\begin{array}{c}-200-2\\ 0-2\\ 8-2\end{array}\right]\mathrm{km}.& \left(38\right)\end{array}$
where estimation errors {circumflex over (r)}−r represents the initial relative navigation errors at the time of target acquisition.

[0145] Two cases are considered: uninterrupted visible sensor measurements and twice-interrupted measurements. For the uninterrupted case, the true relative velocity {dot over (r)} of the chaser in the LVLH frame and its estimate {dot over ({circumflex over (r)} are initialized as

[0146] $\begin{array}{cc}\stackrel{.}{\underset{\_}{r}}=\left[\begin{array}{c}3\omega z_{\mathrm{true}}/2\\ 0.2\\ 0.2\end{array}\right]m/s,\widehat{\stackrel{.}{\underset{\_}{r}}}=\stackrel{.}{\underset{\_}{r}}+\left[\begin{array}{c}0.0\\ 0.2\\ 0.0\end{array}\right]+\Delta \underset{\_}{V}*1.005m/s& \left(39\right)\end{array}$
where 0.5 percent is the ΔV measurement error of the accelerometer.
For the measurement interruption case, the initial values of {dot over (r)} and {dot over ({circumflex over (r)} are slightly different

[0147] $\begin{array}{cc}\underset{\_}{\stackrel{.}{r}}=\left[\begin{array}{c}3\omega z_{\mathrm{true}}/2\\ 0.2\\ 0.0\end{array}\right]m/s,\widehat{\underset{\_}{\stackrel{.}{r}}}=\underset{\_}{\stackrel{.}{r}}+\left[\begin{array}{c}0.0\\ 0.4\\ 0.2\end{array}\right]m/s& \left(40\right)\end{array}$
Apart from the ΔV firings for guidance purposes, there is no other differential acceleration modeled in the simulation. If any differential acceleration exists between two vehicles, it must be modeled and accounted for in the relative motion dynamics.

[0148] The above scenario is implemented using guidance and relative navigation algorithms formulated in Ref. 1 based on Clohessey-Wiltshire equations. FIGS. 7a and 7b illustrate the true and the estimated in-plane trajectories of the chaser satellite relative to the target satellite according to the nominal scenario of FIG. 6 with the initial conditions as specified by Eqs. (38–40). The visible sensor provides angle measurement of the target at 4 Hz. With the first few angle measurements, the relative Kalman filter changes the initial z estimate from 6 km to a {circumflex over (z)} that is proportional to {circumflex over (x)}(0), or {circumflex over (z)}={circumflex over (x)}(z/x)=−202×10/(−195)=10.36 km. Because the relative altitude is almost constant (there is a slight variation in z because ż=0.2 m/s instead of ż=0), until the downrange x reduces to −28 km, the {circumflex over (z)} does not approach the true z (with the relative position estimation errors shown subsequently, in FIGS. 10–11). In FIGS. 7a, 8a, 9a, 10a, and 11a the visible sensor measurements are uninterrupted, while in FIGS. 7b, 8b, 9b, 10b, and 11b the measurements are interrupted twice, each time for an hour, from t=15 to 75 minutes and again from t=95 to 155 minutes. In these figures, the beginning and end of the interruptions are marked by ◯ and □ symbols. In FIG. 7b, because of absence of any unmodeled deterministic acceleration, the absence of measurements does not noticeably deteriorate the estimated trajectory of the chaser. FIGS. 8a and 8b show the true in-plane line-of-sight (LOS) angle measurement, elevation, or pitch, of the target and its measured value with measurement noise of the visible sensor as specified in Table 1. (In this example, the visible sensor focal plane measurements and its attitude have been merged together.) The relative navigation Kalman filter of one embodiment of the present invention uses these elevation and azimuth (not shown here, for brevity) discrete measurements at 4 Hz and estimates the relative motion state vector [x, y, z, {dot over (x)}, {dot over (y)}, ż]^T. From these estimates, the line-of-sight (LOS) angles are estimated. FIGS. 9a and 9b show the estimation error {circumflex over (θ)}_y−θ_y. By comparing the uninterrupted measurement case, FIG. 9a, with the interrupted measurement case, FIG. 9b it may be seen that the pitch angle estimation error grows to 1–2 mrad when angle measurements are not available. But as soon as angle measurements resume at t=4500 s (75 minutes) and 9300 s (155 minutes), these estimation errors drop to the measurement noise level (0.2 mrad, 1σ, for x≧11 km). The x- and z-estimation errors versus time are illustrated in FIGS. 10a and 10b and FIGS. 11a and 11b, respectively. These estimation errors are compared with the square root of the corresponding diagonal elements of the error covariance matrix, ±√{square root over (P_xx)} and ±√{square root over (P_zz)}, respectively. The magnitude of the initial x-estimation error is much larger than √{square root over (P_xx)}, and a gradual decrease in the x-error is observed until the target range becomes measurable (≦10 km) by a laser rangefinder. In FIG. 10b, it is shown that the x-estimation error continues to decrease even in the absence of angle measurements because (a) just prior to the measurement cut-off a good estimate of the state vector [x, y, z, {dot over (x)}, {dot over (y)}, ż] is obtained, (b) there are no unmodeled accelerations in the simulation, and (c) it can be shown that the x-estimation error depends on {tilde over (z)}/z where {tilde over (z)} is the z-estimation error. Therefore, as x decreases linearly, so does the x-estimation error. The element √{square root over (P_XX)} diverges when the measurements are absent, as seen in FIG. 10b, but when the measurements resume it drops to levels achieved with uninterrupted measurements. The {tilde over (z)} estimation error versus time in FIG. 11a does not diminish with angle measurements because z is almost constant and {circumflex over (z)}≈z({circumflex over (x)}/x). When the short-range laser rangefinder becomes operational around t=10,000 s, {tilde over (z)} decays quickly to nearly zero. If the measurements are interrupted, the variance P_zz grows, but, similar to P_xx, it drops as soon as the measurements resume to levels of P_zz without interruption.

[0149] Finally, analogous to the x-estimation error, FIGS. 12a and 12b illustrate the range estimation error versus range with and without measurement interruptions. Consistent with the x-estimation error, the initial range estimation error transitions to values more in line with the angle measurements, measurement noise, and relative translation dynamics. After transients, the range estimate error decreases steadily as range decreases due to the drift at a constant altitude. As before, measurement interruptions in FIG. 12b have virtually no effect on the net range estimation error. The square root of the norm of the covariance matrix of the position vector (x, y, z), √{square root over (P_rr)}, is shown in FIG. 13a for the no interruption case and in FIG. 13b for the interruption case. As with the √{square root over (P_xx)}, the √{square root over (P_rr)} decreases steadily as range decreases when measurements are not interrupted. The √{square root over (P_rr)} grows when angle measurements are interrupted, but, similar to √{square root over (P_xx)}, drops to nominal values when measurements resume.

[0150] These results show that navigating a chaser satellite from a long range (˜200 km) with angle measurements of only a visible sensor has no adverse effect on the relative position estimation errors even if the angle measurements are interrupted twice, each time for an hour.

[0151] The preceding results were obtained treating both vehicles as point masses, assuming zero gyro/star tracker noise, and having no unmodeled acceleration. These limitations are removed in the following illustrations.

[0152] In this section, attitude determination errors arising from gyros and star tracker noises as filtered by the analytic steady-state six-state Kalman filter of one embodiment of the present invention are illustrated. Table 2 records the noise parameters of a SIGI gyro and a selected star tracker. The star tracker noise about its focal plane axes, 35 μrad, (1σ), is one-eighth of the noise about its boresight axis (280 μrad, 1σ). The three nondimensional parameters S_e, S_v and S_u corresponding to the gyro errors and star tracker noise about the focal plane axes for different star tracker measurement intervals are also presented in Table 2.

[0153]

TABLE 2
Comparison of SIGI Gyro and CT-633 Star Tracker Noises
Gyro Noise Parameters
Bias (drift rate)	0.02 μrad/s
Rate of change of bias (σu)	4.0 e-12 rad/s3/2
Random walk rate (σv)	0.75 μrad/s1/2
Quantization as white noise (σe)(readout noise, electronic noise)	1.6 ⁢ ⁢ ( = 5.4 / 12 ) ⁢ ⁢ μrad
Star tracker noise σst
Focal plane axis (1σ per axis)	35 μrad
boresight axis (1σ)	280 μrad
Dimensionless parameters comparing
gyro noise with star
tracker noise (σst = μrad)
star tracker measurement interval = T
readout noise: S3 = σ3/σST = 0.05
Drift rate change in T seconds
s u = T 3 / 2 ⁢ σ u / σ st = { 10 - 7 for ⁢ ⁢ T = 1 ⁢ ⁢ s 6 ⁢ e - 4 for ⁢ ⁢ T = 300 ⁢ ⁢ s
Random walk in T seconds
s v = T 1 / 2 ⁢ σ v / σ st = { 0.02 for ⁢ ⁢ T = 1 ⁢ ⁢ s 0.37 for ⁢ ⁢ T = 300 ⁢ ⁢ s 1 for ⁢ ⁢ T = 2177 ⁢ ⁢ s ⁡ ( ~ 36 ⁢ ⁢ minutes )

The parameter S_e=0.05 indicates that the star tracker is twenty times noisier than the gyro. The noise arising from the change in the gyro drift rate in 1 second is 10⁻⁷ compared to the star tracker noise σ_st and 6×10⁻⁴ in 300 seconds. Lastly, random walk of the gyro in 1 second is 0.02 compared to the star tracker noise and 0.37 in 300 seconds. These parameters suggest that if the star tracker measurements are used frequently (˜1 s) to correct for gyro errors, the attitude estimation error will be of the order of gyro readout noise (1.6 μrad, 1σ), and random walk error will be largely corrected.

[0154] In support of this observation, FIGS. 14a and 14b illustrate roll, pitch, yaw, attitude and drift rate estimation errors before and after star tracker measurement updates. These results are obtained for a star tracker whose boresight axis is in the orbit plane and tilted toward zenith by 15 degrees from the —x-axis. This results in roll, pitch, and yaw star tracker noise equal to 280, 35, and 106 μrad, respectively. This star tracker orientation applies to all remaining example embodiments described below. It may be seen in FIGS. 14a and 14b that the attitude estimation errors of gyros grow significantly due to random walk as the star tracker measurement interval increases, and the drop in the estimation error with the star tracker measurement update is insignificant even when the measurement interval is 100 s because the gyro drift rate change is extremely low. While FIGS. 14a and 14b suggest that the star tracker measurement interval should be as short as 1 s or even shorter, Ref. 3 indicates that, for low-Earth orbits, this interval should be greater than or equal to 6 s so that the low- and high-frequency spatial noises in the star tracker measurements, not modeled here, disappear. More detailed illustration of the influence of star tracker measurement interval on laser rangefinder pointing accuracy is presented below.

[0155] It is demonstrated above in Example Scenario 1 that when a chaser satellite is drifting towards a target satellite, the visible sensor along with a Kalman filter for relative navigation of one embodiment of the present invention yields a satisfactory estimate of range between the two spacecraft, and therefore it is not necessary to measure the distance with a laser rangefinder (LRF). From a distance of 10 km (the maximum range of an LRF), pointing a laser rangefinder at a target requires a tight attitude control that may be expensive to implement. Nonetheless, should a laser rangefinder be used, it is now illustrated that the LRF pointing architecture of one embodiment of the present invention shown in FIG. 1 can point the LRF to a target within the accuracy requirement. The scenario considered is shown in FIGS. 15a and 15b wherein the chaser satellite is 8.8 km behind and 4.8 km below the target, about 10 km away. The initial relative velocity of the chaser satellite is such that it approaches the target satellite at the origin of the LVLH frame along the trajectory in the orbit plane as shown in FIGS. 15a and 15b. The visible sensor measures the azimuth and elevation of the target image in the focal plane, the gyros and star tracker measure attitude of the sensor focal plane, the six-state analytic steady-state filter of one embodiment of the present invention estimates the chaser attitude based on the gyro and star tracker measurements, this attitude estimate and the visible sensor measurements are used as shown earlier in an embodiment of the present invention, and the relative navigation Kalman filter of one embodiment of the present invention estimates the relative position and velocity of the target. The estimated and the true relative position vectors are shown in FIG. 16a and the corresponding estimation errors in x, y, z and the range r are shown in FIG. 16b.

[0156] As expected, the position estimation errors decrease with time, seen clearly for the range error from 600 to 375 meters. Using these estimates and the velocity estimates, the pointing and rate commands of one embodiment of the present invention are calculated as formulated earlier and inputted to the attitude controller. These commands are tracked as illustrated in FIG. 17 for roll, pitch, and yaw rate commands. As a result, after initial transients, the target image remains within the visible sensor focal plane and the azimuth and elevation of the image motion relative to the focal plane center is shown in FIG. 18 when the visible sensor, LRF, gyros and star tracker each are oriented perfectly or near perfectly. FIG. 18 shows that the visible sensor and, therefore, the collimated LRF are pointed at the target well within the pointing accuracy requirement of 88 μrad. In this exemplary embodiment of the present invention, three gyro/star tracker update intervals are considered: 1, 15, and 300 s. Though, contrary to expectations, as the gyro update interval increases, the pointing accuracy improves; this conclusion changes when sensor misalignments are considered.

[0157] Since the LRF pointing is achieved by feeding back the visible sensor angle measurements, it is reasonable to expect reduced sensitivity of pointing accuracy to a misalignment between visible sensor, gyros and star tracker.

[0158] FIGS. 19a and 19b illustrate relative navigation and attitude controls working together to effect laser rangefinder pointing in the presence of both gyros and a star tracker independently misaligned with respect to the visible sensor frame, while the visible sensor and the laser rangefinder are collimated perfectly or near perfectly. The gyro frame is assumed to be orthogonal but misaligned by 2 millirad about each axis in one example and 5 millirad in the other example. Denoting a misalignment angle vector as δθ_ma, the transformation matrix from the perfectly aligned frame to the imperfectly aligned but orthogonal frame is given by [1−δθ_ma^x]. In the two examples, the star tracker, in comparison to the gyros, is misaligned by −2 and −5 mrad about each axis, respectively. In this misalignment scenario, in contrast with the example of perfect alignment in FIG. 13, the star tracker update interval is found to influence the laser rangefinder pointing accuracy significantly, and an interval of 1 s is determined to be essential to meet the pointing accuracy with margin, as illustrated in FIGS. 19a and 19b. FIG. 19a compares the target image motion in azimuth for ±2 and ±5 mrad misalignments; FIG. 19b compares the target image motion in elevation for ±2 and ±5 mrad misalignments. The laser rangefinder pointing accuracy is, after initial transients, within the pointing requirement of ±88 μrad.

[0159] FIG. 20 illustrates the LRF pointing accuracy for gyro and star tracker misalignment angles reduced to 1 and −1 mrad/axis, respectively, but with additional 0.5 mrad/axis misalignment of the visible sensor and 0.55 mrad/axis misalignment of the LRF with respect to their respective ideal orientations. In other words, the LRF and the visible sensor boresights are misaligned by 50 μrad about azimuth and elevation axes. The 50 μrad bias error is seen in target image motion in the visible sensor focal plane in FIG. 20 that compares the star tracker measurement update intervals of 1 and 15 s. In FIG. 20, apart from seldom exceedancee in azimuth of ±88 μrad pointing requirement, the LRF pointing accuracy is within requirements even in presence of the above selected misalignment angles of the four sensors.

[0160] Table 3 summarizes the allowable misalignments of the four sensors. These misalignments are unknown to the laser rangefinder pointing technology embodiment of the present invention presented herein. If the misalignment between an LRF and a visible sensor can be controlled to a greater precision during installation, 20 μrad instead of 50 μrad for instance, then larger (than 1 mrad) misalignments of gyros and star tracker will be permissible.

[0161]

TABLE 3
Limiting Misalignment Angles of Various Sensors
	Gyro	<1 mrad
	Star tracker	<1 mrad
	Visible sensor	<0.5 mrad
	Visible/LRF	<0.05 mrad

[0162] A high-fidelity model of reaction wheel speed/torque characteristic is implemented in the simulation of exemplary Scenarios 2 and 3 above. See Ref. 13.

[0163] The preceding observability analysis and simulation results indicate that angle measurements of a visible camera may be used effectively for gross estimation of range of a chaser satellite from a target satellite. The range estimation error does not become zero rapidly with angle measurements, but decreases gradually as the chaser satellite approaches the target satellite. In the event of visible sensor measurement interruptions, range estimation errors do not grow; instead, they continue to decrease, provided that before interruption, a relative navigation filter has minimized its steady-state relative position and velocity estimation errors and there are no relative accelerations unmodeled or unknown to the filter. In view of these observability results, a rendezvous sensor suite of one embodiment of the present invention may not include a ˜10 km laser rangefinder and an associated tight-pointing controller.

[0164] If a laser rangefinder is nonetheless used to determine the relative range between the two satellites, a tight pointing control system for the laser rangefinder of one embodiment of the present invention is provided that includes visible sensor focal plane angle measurements, an extended Kalman filter for relative navigation, pointing and rate commands, a six-state attitude determination Kalman filter using gyros and a star tracker, and an attitude controller. The laser rangefinder pointing architecture of one embodiment of the present invention achieves the estimated long-range pointing accuracy requirements. This architecture of one embodiment of the present invention is useful in other flight phases as well; for instance, it is effective in a 24-pulse closed-loop circumnavigation of the target satellite at a radius of 100 meter.

[0165] FIG. 21 is a flowchart of a method for a system of a closed-loop autonomous guidance, navigation, and control architecture of one embodiment of the present invention. The flowchart presents subsystems and data calculations for the inertial navigation component, relative navigation component, and pointing and rate commands component with the associated devices required or used with these different subsystems. External objects, such as GPS satellites, constellation 102 and the target satellite 128 are provided for reference to show where data can be captured from these external objects by the system 100. For example, in the inertial navigation component, the SIGI, or space integrated GPS and INS system, 110 acquires information and data measurements from the GPS satellites constellation 102 using a GPS receiver 118. Similarly, a laser range finder 132, visible sensor 134, or infrared sensor 136 of the rendezvous suite 130 obtains measurement data from the target satellite 128. The inertial navigation component of the embodiment includes the SIGI system 110, a star tracker 124, and an inertial attitude Kalman filter 126. The relative navigation component includes a rendezvous suite 130 and a relative navigation Kalman filter 140. The pointing and rates commands 142 include quaternion pointing commands 144 and rate commands 146.

[0166] The flowchart presented in FIG. 21 presents a summary depiction of data and measurements that flow between the systems and devices of an embodiment of the present invention. By way of example of the flow of data or measurement data and calculations through the system 110, a GPS receiver 118 provides position and velocity of the chaser satellite at time t, gyros 114 provide spacecraft attitude and rate, and accelerometers 116 provide incremental velocities for a SIGI inertial navigation Kalman filter 120. The inertial attitude Kalman filter 126 is presented with gyro 114 output and star tracker 124 measurements. The inertial attitude Kalman filter 126 feeds back attitude correction to the SIGI inertial navigation Kalman filter 120. A visible sensor 134 and an infrared sensor 136 provide azimuth and elevation angle measurements and, possibly, passive range measurements 138. These measurements and the measurements 138 from a laser range finder 132, accelerometers 116, and the SIGI inertial navigation Kalman filter 120 provide input to a relative navigation Kalman filter 140. The relative navigation Kalman filter 140 produces relative position and velocity estimates that are used to generate pointing and rate commands 142 and guidance of the satellite 152. The pointing and rate commands 142 are used for attitude control using reaction wheels 154. The guidance 152 provides incremental velocity commands for translation control of the spacecraft. Together with FIGS. 1, 2A, and 2B, one skilled in the art will recognize application of embodiments of the present invention for relative navigation, attitude determination, pointing, and tracking for spacecraft rendezvous.

[0167] The inertial navigation, relative navigation, and pointing and rate commands components of the present invention are described herein by their separate component elements; however, they are meant to include both the separately described component elements and associated computer programs or computer program products required for performing the described functionality of the component elements.

[0168] An embodiment of the present invention may be effected by a computer system. In this regard, such a computer system may include at least one control processor, such as a processor or central processing unit (CPU), interoperably connected to, and implementing software and data stored by, memory such as random access memory (RAM) or other storage devices. The control processor may interface with other system components to accept inputs and generate outputs, including translation and rotation commands for a chaser satellite to rendezvous with a target satellite. For example, a control processor may interface with system components of inertial navigation, relative navigation, and pointing and rate commands components, a guidance controller, an attitude controller, a reaction jet controller, and a reaction wheel controller. One skilled in the art will recognize that embodiments of the present invention may be one computer system or a group of distributed computer systems. One or more processors, memory, storage devices, and other computer elements may be used in common by a computer system and subsystems, as part of the same platform, or processors may be distributed between a computer system and subsystems, as parts of multiple platforms.

[0169] It will be understood that each block, step, or element of the flowcharts described herein, and combinations of blocks and/or elements in the flowchart described herein, support combinations of means and combinations of steps for performing the specified functions. Similarly, it will be understood that each block, step, or element of the flowchart, and combinations of blocks and/or elements in the flowchart, can be implemented by computer program instructions. These computer program instructions may be loaded onto the computer system described above or other programmable apparatus to produce a machine, such that the instructions which execute on the computer or other programmable apparatus create means for implementing the functions specified in the flowchart block(s) or element(s). These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart block(s) or element(s). The computer program instructions may also be loaded onto a computer or other programmable apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block(s) or element(s). It will also be understood that each block or element of the flowchart, and combinations of blocks and/or elements in the flowchart, can be implemented by special purpose hardware-based computer systems which perform the specified functions or steps, or combinations of special purpose hardware and computer instructions.

[0170] Many modifications and other embodiments of the inventions set forth herein will come to mind to one skilled in the art, to which these inventions pertain, having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. Therefore, it is to be understood that the inventions are not to be limited to the specific embodiments disclosed, and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.