1. Mars Science Laboratory Navfilter Trajectory Reconstruction Fred Serricchio Miguel San Martin, Edward C. Wong Jet Propulsion Laboratory, California Institute of Technology IPPW San Jose, CA June 17 - 21, 2013 (c) 2011 California Institute of Technology. Government sponsorship acknowledged. Copyright 2011 California Institute of Technology. Government sponsorship acknowledged.

2. Outline GNC Challenges Navfilter Architecture (IMU, TDS) Performance Velocity Analysis Ground Estimates Attitude Initialization TDS Performance Conclusions/Open Items 9/13/07 2

3. MSL EDL/GN&C Challenge EDL Challenge Land a 950Kg Rover over terrain with: –< 30 degree slopes, and –< 1.2 meter rocks Response 6-Axis Powered-Descent + SkyCrane –Parachute terminal velocity Vertical = 65 to 100 m/sec Horizontal = 0 to 40 m/sec –Powered-Descent Start Altitude 1.5 to 2 km –Rover touchdown velocity Vertical = 0.75 m/sec, +/- 0.1m/sec (3-sigma) Horizontal = 0 m/sec, +/- 0.5 m/sec (3-sigma) –Descent Stage touchdown velocity Vertical = 0.75 m/sec, +/- 0.1 m/sec (3-sigma) Horizontal = 0 m/sec, +/- 0.1 m/sec (3-sigma)

4. NavFilter Architecture Requirements –Attitude Initialization Original.25 Deg 3σ, Reduced to 0.1 Deg 3σ before landing –Touchdown Velocity Knowledge Error 0.1 m/s 3σ all axes Provide state information to all modules throughout EDL Inertial Propagation –200 Hz (DIMU sample rate) integration of Attitude, Position, and Velocity –DIMU internal Coning and Sculling compensation (Internal DIMU sampling rate is 1600Hz) –J2 Gravity Model Attitude Initialization from Cruise ACS Position and Velocity state Initialization from Ground Navigation in EdlParms (Epoch at E-9minutes) 9/13/07 4

5. IMU 2 IMUs (primary and backup) Honeywell MIMU It provides 3 axis position and attitude knowledge in the form of: Accumulated angle for each DIMU axis. Accumulated sensed acceleration for each DIMU axis. Operation: One IMU operational at a time. IMU selection done before EDL. DIMU (MIMU) http://www51.honeywell.com/aero/common/documents/myaerospacecatalog-documents/MIMU.pdf

6. Terminal Descent Sensor The Radar Terminal Descent Sensor produces line of sight range and velocity measurements on 6 independent narrow (~3.5 deg) beams –Single antenna per beam, with 3 beams canted 20 o off nadir, 2 beams canted 50 o off nadir, and 1 nadir beam –Ka-band (35.75 GHz) center frequency –Pulse doppler radar design The Radar Terminal Descent Sensor produces line of sight range and velocity measurements on 6 independent narrow (~3.5 deg) beams –Single antenna per beam, with 3 beams canted 20 o off nadir, 2 beams canted 50 o off nadir, and 1 nadir beam –Ka-band (35.75 GHz) center frequency –Pulse doppler radar design

7. Descent Stage Configuration Terminal Descent Sensor (TDS) 1x Descent IMU (DIMU) 2x Mars Landing Engines (MLEs) 8x Entry RCS 8x Descent Stage Entry Configuration Descent Stage Skycrane Configuration Terminal Descent Sensor (TDS)

8. NavFilter Architecture (cont.) TDS Description –Ka-band (35.75 GHz) center frequency –Pulse doppler radar design, LOS Velocity and LOS Slant Range –6 beams: 3 beams canted 20° off nadir, 2 beams canted 50° off nadir,1 nadir beam TDS data is used to correct errors in Inertially propagated Velocity, Acceleration and Altitude Velocity Filter is 6 state Least Squares Filter –3 Axis Velocity Error –3 Axis Acceleration Error (attitude, misalignment, etc) –Recursive (but finite) data editing Altitude Filter –Estimates ground over MSF frame –Mean of ground altitudes in measurement buffer 9/13/07 8

9. NavFilter Architecture Inertial Propagation (64 Hz) 9/13/07 9 Attitude Propagator Attitude Propagator DIMU 200 Hz DIMU 200 Hz Velocity Propagator Velocity Propagator Position Propagator Position Propagator Coordinates Frames & Center of Navigation Transforms Coordinates Frames & Center of Navigation Transforms Position Correction Propagator Position Correction Propagator V corr (t i ) + A corr (t i ) x ( t – t i ) ∆h 0 V corr (t i ) dimuAtt_q_Dimu_J + + + - - - dimuPos_J dimuVel_J deltaV_Dimu deltaTheta_Dimu From Cruise: dimuAtt_q_Dimu_J(0) From Ground Nav: Pos_J(t 0 ), Vel_J(t 0 ) To TDS Batch Filter (Foreground) To PD Controller q_Msf_M PosMsf_M Mars Rotation Mars Rotation q_M_J, wMars_J wMars_J RotRate & Acc Filter RotRate & Acc Filter deltaTheta_Dimu, deltaV_Dimu w_Ds, wDot_Ds, acc_Ds uncorr_dimuPos_J uncorr_dimuVel_J dsPos_Msf dsVel_Msf q_Ds_J q_Msf_J

10. NavFilter Architecture Measurement Update 10 TDS 20 Hz TDS 20 Hz Sanity Checks Sanity Checks Compute Velocity Residual Compute Velocity Residual Compute Raw Ground Altitude Compute Raw Ground Altitude Foreground Buffer (1 sec) Foreground Buffer (1 sec) Data Editing Data Editing Final Fit Final Fit Validity Checks Validity Checks Ground Altitude Rectification & Form Solution Buffers Ground Altitude Rectification & Form Solution Buffers Data Editing Data Editing Data Editing Data Editing Final Fit Final Fit Final Fit Final Fit Validity Checks Validity Checks Validity Checks Validity Checks Primary Solution Secondary Solution Final Solution Selection Final Solution Selection Background Buffer (20 sec) Background Buffer (20 sec) timeTag, beam_ID, vel_residual, beam_direction_J timeTag, beam_ID, raw_ground_alt_inMsf t epoch, V corr (t epoch ), A corr (t epoch ) ground_alt To Inertial Propagator To PD Controller Forward (64 Hz) Background (1 Hz) Form Solution Buffer Form Solution Buffer

11. Flight Reported Performance Flight Reported Values Velocity Correction when Navfilter converged [.5884 -.4567.1942 ] m/s (MSF Frame, East-North-Up) Velocity correction.77 m/s Altitude Correction when Navfilter converged 113.4 meters Touchdown Velocity [ -.02203 -.003505 -.76501 Z ] m/s ( MSF Frame, East-North-Up) Position of Descent Stage Reported at Touchdown [ 1597.74 246.3 7.677 ] meters ( MSF Frame, East-North-Up ) Knowledge Error at Touchdown Based on MRO photos Using OD179770 meters ( Flight, 5 day old at time of EDL) Using OD230162 meters ( After landing rerun standalone Navfilter using od230 state) The Navfilter converged at an altitude of 8500 meters –Needed to converge by 2500 meters to accomplish EDL activites before safe touchdown

12. Flight Reported Velocity Solutions The first thing we noticed was the acceleration term in the velocity solution for Z axis Acceleration of 450 micro G’s

13. Local Gravity At Gale Crater We were able to match the slope in the velocity solutions we saw in flight That resulted in a larger velocity Eastward at touchdown than reported landing night Reconstructed Velocity

14. Touchdown Velocity Analysis Re-Run the Navfilter without any TDS data after 935 seconds past T0 (~450 m above terrain) –This was done to remove any potential sandy radar effects on TDS measurements close to the ground –This matched the touchdown velocity from the RIMU (Rover IMU) very well Descent Stage Velocity at Touchdown [.1694 -.0482 -.6312 ] m/s in MSF frame Versus Desired [ 0 0 -.75 ] m/s, +/- 0.1 m/s 3σ

15. Msf: Mars Surface Frame Mars Surface Frame (Msf) –Origin: Intersection of s/c position vector at first Nav. Filter TDS solution with Mars Spherical model with radius equivalent to radius at landing target (or radius at edl_update_param?) –Orientation: ENU (East, North, Up) > East: > North: > Up: along radial vector from center of Mars. AGL altitude Ground altitude Mars Sphere (Radius = radius at landing target) SC position vector Msf Power approach target altitude = + dh_const_decel + dh_const_vel_accordion + h_sky_crane_start Msf frame defined at first Nav Filter Solution using TDS data

16. Ground Estimate We see some vehicle dynamics and ground effect but once we converged the estimates of the ground were not changing that much 10 m ~70 m

17. Attitude Initialization Error Analysis –Using best estimate of reconstructed trajectory to determine the best estimate of gravity using both RIMU and DIMU data –Obtain the best estimate from gravity at touchdown location from the Navigators using MGM2011 knowledge Best Estimate of Attitude Init error is 30 millidegrees

18. TDS Performance –The TDS produced no ambiguities during all of EDL –The Navfilter Data Editing was never invoked –TDS passed 1 measurement of the Heat shield and the Navfilter rejected it in the pre-filter

19. Conclusions/Open Items Performance –Attitude Init. 30 millidegrees –Position Knowledge error 162 meters –Velocity at touchdown [.1694 -.0482 -.6312 ] m/s –Velocity Knowledge error [ -.19.045 -.134 ] m/s (almost 5σ) Open Items –Sandy radar effects and the near ground effects of radar