using onboard telemetry for maven orbit determination...maven background •mars orbiter studying...

AAS/AIAA Space Flight Mechanics Meeting, 14.Jan.2014 1

Using Onboard Telemetry for

MAVEN Orbit Determination

Drew Ryan Jones

Try Lam, Nick Trawny, Cliff LeeJet Propulsion Laboratory, California Institute of Technology

2AAS/AIAA Space Flight Mechanics Meeting, Jan-14-2014

MAVEN Background

• Mars orbiter studying the upper atmosphere and

exploring the climate history

2

• Science Orbit:

– 4.5 hour period

– 75 degree inclination

– Range of periapsis latitudes

– Periapsis altitude ~140-170

km (<120 km Deep-Dips)


• MAVEN Navigation requirements:

– Target a density corridor (0.05-0.15 kg/km3)

– Predict periapsis passage to 20 seconds

• Navigation challenges:

– Limited Doppler tracking (only 1-in-5 orbits)

– Between aerobraking and prior Mars science orbits

– Frequently changing attitude (incl. drag passes)

– Momentum desaturations after every periapsis (+11 min)


Modeling

• Mars-GRAM 2005 (MG05) - baseline atmosphere,

only estimate scale factor

– Quality density reconstruction better prediction

– But Mars atmosphere does not play nice!

• Lat/Lon dependence, unexpected trends, ect

• Scale factor variance up to 100% (orbit-2-orbit)

• Estimated Parameters (baseline):

– Epoch state

– Solar radiation pressure scale factor

– MG05 scale factor as stochastic (batch update each orbit)

– Desats as impulsive burns

4


Motivation / Goals

• Supplement sparse radiometric data with telemetry

derived measurements

– Improve density reconstruct/predict performance

– Combine observables into single batch estimation

• Generally valuable capability

– Enable cheaper (less tracking) and autonomous missions

– Applications to non-grav estimation:

• Aerobraking

• Thruster and SRP calibrations

• Finite burn reconstruction

6


IMU Observable Formulation

• Accelerometer (IMU):

– IMU frame body frame via gyros (onboard)

• CFD model of tabulated drag/moment coefficients

– Interpolant is relative velocity direction on unit sphere

Typically just compute bias (and IMU clock offset) apriori, and don’t

estimate further

7


Results: MAVEN cruise (TCM-1)

8

• Largest maneuver

until MOI

• 6-second IMU

clock offset

observed


Results: MAVEN cruise (TCM-1)

9

Fitting offset IMU data with

Doppler data does not work


Results: MAVEN Science

• Single drag pass fit (F2 and IMU)

– Bias of around 5 μm/s2

– Peak IMU acceleration: 100 μm/s2

– Centroid shifted by ~30 seconds

IMU observed net ΔV:

8.6 mm/s

Scaled MG05 net ΔV:

10.8 mm/s


Typical MAVEN

reconstruction batch


Conclusions

• Demonstrated uses:

– Reconstructed density information for 4/5 orbits where

radiometric data is absent

– Atmospheric functional data (beyond net ΔV)

• Larger ΔV during first deep-dip will be a real test

• Absence of telemetry during desats unfortunate

– If available, would it help? Noise still prohibitive?

• Further work

– Reduce data noise in IMU measurements (smoothing)

– Use alternate atmosphere model (more parameters)

– Utilize/test torque observable


Questions?

[email protected]


BACKUP


Torque Observable Formulation

• CFD model of tabulated drag/moment coefficients

– Interpolant is relative velocity direction on unit sphere

• Torque from reaction wheel speeds

15


Results: MAVEN science

• Drag pass in IMU observable (unsmoothed)



16


Results: MAVEN Science

• Single drag pass fit (F2 and IMU), unsmoothed



– Centroid shifted by ~45 seconds

IMU observed net ΔV:

9.1 mm/s

Scaled MG05 net ΔV:

10.8 mm/s


IMU accelerations, orbit 307

IMU accelerations, orbit 314

MAVEN reconstruction batch

• IMU data suggests less ΔV

• IMU centroid always later (but

as little as 5 sec)

IMU observed

net ΔV: 4.1

mm/s

Scaled MG05

net ΔV: 6.0

mm/s

IMU observed

net ΔV: 3.4

mm/s

Scaled MG05

net ΔV: 4.6

mm/s


MRO: Mars-GRAM Scale Variation

using onboard telemetry for maven orbit determination...maven background •mars orbiter studying...

Documents