Hayabusa2 Landers

DLR. CC BY 3.0.

Overview

Hayabusa2 carried three small surface vehicles to asteroid 162173 Ryugu [1]: the twin MINERVA-II1 rovers Rover-1A and Rover-1B, built at JAXA/ISAS, and MASCOT, built by DLR with CNES. All three are hoppers. MINERVA-II is the program name, successor to the single MINERVA carried by Hayabusa; MINERVA-II1 is the first of its containers, the one holding Rover-1A and Rover-1B.

Specifications

Parameter	MINERVA-II1 Rover-1A / 1B	MASCOT	Source
Body	18 cm diameter, 7 cm high	30 x 30 x 20 cm	[1], [7]
Mass	1151 g / 1129 g	about 10 kg	[1], [7]
Mobility	hopping, single DC motor driving a flywheel	hopping, eccentric tungsten swing arm driven by a stepper motor	[1], [7], [5]
Hop performance	about 10 cm/s for a 1 kg rover in microgravity, from simulation and drop tower test	up to about 100 m in a single hop	[1], [5]
Power	solar cells on every face, 2 W maximum	none; 200 Wh primary battery sized for 16 h	[1], [7]
Energy storage	3 electric double layer capacitors in series, charged above 6 V, computer stops below 4 V	primary battery only	[1], [7]
Processor	44 MHz, no floating point unit, 256 kB program limit	not published	[1]
Telemetry	32 kbps maximum to the spacecraft relay, over 20 km range	not published	[1]
Cameras	4 (Rover-1A), 3 (Rover-1B)	MasCam, plus MARA, MicrOmega and MasMag	[1], [2]

Mission profile

Parameter	Value	Source
Host	Hayabusa2	[1]
Target	asteroid 162173 Ryugu	[1]
MINERVA-II1 separation	21 September 2018 at 04:06 UTC, from about 50 m altitude, the spacecraft translating north so that its own motion canceled the separation velocity	[1]
MASCOT separation	3 October 2018, from about 51 m	[7]
MASCOT time to first contact	about 6 min	[7]
MASCOT achieved lifetime	17 h 07 min, about two asteroid day and night cycles	[7]
MASCOT final resting position	22.32 +/- 0.05 deg S, 317.16 +/- 0.05 deg E	[2]

Operating environment

Quantity	Value	Source
Volume-equivalent diameter	850 to 880 m	[3]
Rotation period	7.6326 h	[3]
Day and night at the surface	3.8 h each	[1]
Surface gravity	about 1e-4 m/s2, about 10 micro-g	[1]
Local gravity at the Rover-1A site	1.3e-4 m/s2 from the shape model	[1]
Modeled surface acceleration used pre-encounter	1.8e-4 m/s2	[3]
Escape velocity	about 35 cm/s	[1]
Global thermal inertia	225 +/- 45 J/m2/K/s^0.5	[6]
Geometric albedo	0.049	[3]

The pre-encounter prediction from remote thermal inertia of 150 to 300 J/m2/K/s^0.5 put the typical regolith particle size at 3 to 30 mm, with 6 to 10 mm most likely [3]. With an assumed 50 percent porosity, lunar-analogue relations give a cohesion near 300 Pa and a friction angle near 37 degrees, falling to perhaps 100 Pa if the grains are 1 mm to 1 cm. That set of numbers, not a fine dust regolith, is what the landers were designed against.

Wheels are not usable here. Available traction scales with normal force, which scales with weight, and at 1e-4 m/s2 the weight of a 1 kg rover is on the order of 0.1 mN; cohesion between surface grains exceeds their weight across the whole relevant size range [8]. All three vehicles therefore react internally rather than against the surface.

MINERVA-II1 rovers

Beyond the parameters in the comparison table above:

Parameter	Value
Navigation sensors	photodiodes, thermometers, gyro, accelerometer
Observation sensors beyond the cameras	thermometers, potentiometers

Source: [1].

Hopping mechanism

A single internal DC motor drives a flywheel. Spinning it up generates a reaction torque on the body, and because the point of application of the force between rover and surface is offset from the center of mass, friction converts part of that torque into a lateral force, so the rover leaves the surface with horizontal velocity [1]. Larger normal force gives larger friction for a short time and therefore a faster hop. Numerical simulation and drop tower testing give about 10 cm/s for a 1 kg rover in microgravity [1].

The motor must be turned while the rover is at rest on the surface. Firing it in free flight only changes attitude and produces no translation [1]. The friction the method depends on is itself uncertain at this gravity, because interparticle cohesion rather than weight sets the shear response of the surface layer [8], and the pre-encounter estimate spanned 100 to 300 Pa of cohesion with a friction angle near 37 degrees [3].

Flight results for the Rover-1A hop of 22 September 2018, at about 30 degrees south [1]:

Quantity	Value
Motor drive	75 percent duty
Time from hop to return, from gyro history	1024 +/- 7.5 s
Assumed initial speed	6 to 8 cm/s
Assumed hop angle from drop tower results	30 to 45 degrees from vertical
Predicted flight time for that range	700 to 1100 s
Ballistic range	25 to 50 m

Rover-1B hopped on asteroid day 7, 23 September 2018, imaged the asteroid six times from above during the flight, and returned to the surface 24 minutes later, bouncing several times before coming to rest [1]. Multiple bounces on arrival are the expected behavior for a body whose surface is a poorly sorted boulder field rather than a fine deposit [2].

Autonomy

The rovers cannot localize themselves. The processor has no floating point unit and cannot evaluate imagery to determine rover state, so state estimation runs on the photodiodes or the gyro instead [1]. A rover at rest on the surface has near-zero angular rate and a photodiode signal that does not change; one in ballistic flight has a non-zero angular rate, because after a hop it necessarily carries angular momentum, and its photodiode readings vary as it tumbles. The estimator classifies the rover as in ballistic flight or at rest, and only in the second state does it fire the motor, image in stereo, and measure surface temperature and surface potential relative to the rover.

Because images during flight are taken at random attitudes, most contain no asteroid. Onboard screening rejects them before transmission [1]. Uncompressed images are screened by four tests on the luminance histogram of a decimated Y channel: too many pixels near 0 is a black image, too many near 255 is a white-out, and a luminance variance below threshold is a featureless image. All thresholds are settable by command from Earth. JPEG compression is done in software in spare processor time and takes 6 to 7 minutes per image [1], so the initial configuration transmits uncompressed; when compression is used, an insufficient size reduction is itself the rejection criterion.

Three further autonomous functions run on state: protection in a high temperature environment, transition to a low power mode, and transmit power control [1]. Nothing in this chain depends on knowing where the rover is on the body, which is the design response to a surface whose shape model was only built after arrival [3].

Duty cycle and thermal

The rovers only work in daylight. The capacitors charge after sunrise, the computer starts, and as evening approaches the power balance goes negative, the capacitor voltage falls and the computer stops until the next sunrise [1]. This repeats on the 7.6 h asteroid day.

A protection function cuts power to the equipment above 80 C; it was never triggered [1]. The pre-encounter thermal model put Ryugu at a geometric albedo of 0.049 with an infrared emissivity of 0.9, which is what drove the expectation of a hot surface [3]. Ryugu was expected to run hot because its albedo is very low and it absorbs nearly all incident sunlight, but the short rotation period keeps the surface from reaching that level, and the highest instrument temperature recorded was 45 C on 16 October 2018, asteroid day 82 [1]. On the same day the autonomous voltage watchdog switched the transmitter off late in the day, which restored the power balance and let the voltage recover before the transmitter was restarted.

Results

Rover-1A kept communicating with the spacecraft for 113 asteroid days, until 26 October 2018, and returned 609 images [1]. Rover-1B was last heard on asteroid day 10, 24 September 2018, having returned 39 images. The cause of loss is not established for either, but Rover-1B’s capacitor voltage collapsed immediately beforehand, which is consistent with it having entered permanent shadow. Weak signals earlier in the mission are attributed to the rovers’ own autonomous transmit power control.

MASCOT

Beyond the parameters in the comparison table above:

Parameter	Value
Structure	carbon fiber reinforced plastic laminate, 0.125 mm thick
Structure mass	450 g

Source: [7].

The eccentric swing arm is both the hopping actuator and the uprighting mechanism, which matters because the instruments face one direction [7]. It is driven by a stepper motor and can move the lander up to about 100 m in a single hop [5]. There is no solar array; lifetime is fixed by the primary battery.

Payload and what it measured

MARA is a six-channel thermal infrared radiometer covering about 10 to 15 cm2 within the MasCam field of view [2]. It operated from separation through the whole 17 h mission and, after landing, measured the emitted flux of one rock over a full 7 h 38 min asteroid rotation. Two channels are dedicated to low night-time temperatures, an 8 to 12 micrometer bandpass and a long pass above 3 micrometer [2]; four further bandpasses at 5.5 to 7, 8 to 9.5, 9.5 to 11.5 and 13.5 to 15.5 micrometer resolve emissivity against wavelength. The six channels are designated SiLP (3.5 to 100 micrometer), W10 (8 to 12), B06, B08, B09 and B13 [4].

MARA result	Value	Source
Surface temperature range at the landing site	210 to 300 K	[2]
Rock thermal inertia, first analysis	282 (+93 / -25) J/m2/K/s^0.5	[2]
After data assimilation re-analysis	295 +/- 18 J/m2/K/s^0.5	[2]
After shape model and MASCOT self-reflection correction	256 +/- 34 J/m2/K/s^0.5	[2]
Derived rock porosity	0.46 to 0.56	[2]
Favored porosity from Ryugu bulk density constraints	0.50 +/- 0.02	[2]
Best-fit thermal inertia in the final MARA analysis	256 (+4 / -3) J/m2/K/s^0.5	[4]
Best-fit porosity in the final MARA analysis	46.7 (+0.3 / -0.4) percent	[4]
Broadband emissivity	0.97 +/- 0.01	[4]
Emissivity, 5.5 to 7 micrometer	0.87 +/- 0.01	[4]
Emissivity, 8 to 9.5 micrometer	0.98 +/- 0.01	[4]
Implied grain density	2848 +/- 152 kg/m3	[2]

There was no evidence of a masking dust layer, so the low thermal inertia is a property of the rock itself rather than of a coating [2]. Emissivity across the MARA channels matches the most aqueously altered carbonaceous chondrites, CI, CM and CR, which places Ryugu as strongly aqueously altered before dehydration [4].

MasCam uses a Scheimpflug optic so that the whole scene from 150 mm to infinity is in focus, and carries 4 by 36 monochromatic LEDs in four spectral bands for night imaging [2]. Surface image scale reaches 0.1 mm per pixel. It resolved two boulder types, brighter smooth ones with non-dendritic fractures and darker cauliflower-textured ones consistent with thermal fatigue, with sizes from a few millimeters to a few tens of meters and poor sorting. No accumulation of fine dust was found anywhere in the images. Shadow tracing on the LED-lit night images gives the rock surface a fractal dimension of 1.16 +/- 0.04 and an RMS slope of 36.6 +/- 1.4 degrees, about 6 percent less rough than material imaged on comet 67P [2].

MasMag is a triaxial fluxgate magnetometer.

MasMag parameter	Value
Sensor mass	89 g
Electronics mass	180 g
Total mass	about 280 g
Power	about 0.5 W
Electronics envelope	109 x 94 x 20 mm
Dynamic range	+/- 12,000 nT
Noise at 1 Hz	below 15 pT/sqrt(Hz)
Resolution	2 pT
Sampling rate	10 Hz
Data rate	1010 bit/s
Excitation frequency	about 9.6 kHz

Source: [5].

It acquired data from the start of descent, through the bounces, at touchdown and at the final resting position, and after calibration and de-spinning showed no detectable change above background: the field vector fluctuates within +/- 1 nT with no systematic trend on approach down to about 10 cm from the surface [2].

MicrOmega is a near-infrared hyperspectral microscope intended to determine the mineralogical composition of the regolith, but MASCOT came to rest on a depression that prevented it from returning its intended measurement, and an identical instrument was subsequently applied to the returned samples in the curation facility instead [2], [7].

Site context

The MASCOT landing area was characterized from orbit by Hayabusa2. Regional average thermal inertia there is 181 +/- 102 J/m2/K/s^0.5 and the individual rock MASCOT rested against gives 180 (+65 / -40) [6], against a global average of 225 +/- 45 and a thermal infrared imager estimate of 200 +/- 7. Multi-band orbital imagery of the region depends on image registration across the ONC-T filter set, which is a separate processing problem for a rotating irregular body [10].

Surface strength at depth was constrained independently by the Hayabusa2 Small Carry-on Impactor experiment, whose crater is reproduced by numerical simulation only for very weak material [9].

Radiation

No lander-specific radiation design data is published for either vehicle. The interplanetary environment over the four-year cruise is that of the Badhwar-O’Neill galactic cosmic ray model [13].

Technologies developed

MINERVA-II1 demonstrated that a torque-reaction hop is a workable mobility method at 1e-4 m/s2, and that a 44 MHz processor with no floating point unit can support useful autonomy if the state estimator is built on photodiodes and a gyro rather than on imagery [1]. The image screening filter is the same idea applied to the downlink: a rover that cannot point its camera usefully instead discards its own bad frames.

MASCOT demonstrated an instrumented, uprighting, battery-only lander at asteroid gravity, and produced the only in-situ thermal and magnetic measurements made on an asteroid surface [2]. Hayabusa2 itself continues in extended mission, with a flyby of near-Earth asteroid 98943 Torifune [11], and the mission remains the reference for small-body surface delivery [12].