NASA GRC Simulated Lunar Operations Laboratory

NASA/GRC/ Bridget Caswell. Public domain (NASA / US government work).

The Simulated Lunar Operations laboratory at NASA GRC is NASA’s terramechanics facility, and the anchor of the center’s mTRAX Planetary Exploration group in the Engine Research Building: three simulant soil tanks including a hydraulically actuated tilt bed, a drawbar pull lane, a 777 m3 motion capture volume, and alongside them the TREC single-wheel dynamometer, the Excavation Lab, the XCEL and XTERra tire rigs and the VOiD vacuum dust chamber [1][8]. Its distinguishing asset is not the bins but the terrain preparation standard that goes with them, a published procedure that produces three repeatable GRC-1 conditions whose cone index gradients bracket the values measured on the Moon by Apollo 15 and 16 [2].

Laboratory data

Parameter	Value
Operator	NASA GRC, mTRAX Planetary Exploration team
Location	Cleveland, Ohio, United States [1][8]
Commissioned	Not published; GRC-1 characterized 2010
Type	Terramechanics, excavation and tire-environment laboratory
Floor area	Not published
Capabilities, terrain	GRC-1 lane, tilt bed, Sink Tank, DRAGON M
Capabilities, rigs	TREC, Excavation Lab, XCEL, XTERra, VOiD
Simulant or terrain	GRC-1, GRC-3, GRC-3B, Fillite, #90 silica sand, JSC-1A, pea gravel; tonnages not published [2][8]
Instrumentation	OptiTrack, 32 cameras, sub-mm; Rimik CP40II cone penetrometer
Ground truth	Cone penetrometer at 8 points per test; motion capture pose independent of odometry [1][2]
Fidelity limits	1 g, Earth ambient; GRC-1 cohesionless against up to 1 kPa lunar cohesion [2]
Access	Open to industry, government and academia through the ERB facility manager; no lead time published
Cited by	VIPER [10], Scarab [3][4], Spirit [1]

The laboratory is indoors and climate controlled, and the operators state that stable temperature and humidity are what limit variability in soil conditions between tests. Full scale vehicles, subsystems and components are all accepted, and the material in any tank can be removed and replaced with alternate granular media on request [8]. No open document gives a floor area, an article mass limit, or the tonnage held in any tank. Glenn provides its ground test facilities to industry, government and academia, with the Engine Research Building facility manager and the Test Facility Management Branch published as the route in; no fee schedule is published.

Capabilities

GRC-1 lane

NASA/GRC/Bridget Caswell. Public domain (NASA / US government work).

Parameter	Value
Working volume	12 x 3 x 0.3 m soil tank
Test article limits	Largest run is the 400 kg Scarab rover [3]
Illumination	Not published; no solar simulator
Simulant or terrain	GRC-1, 0.3 m bed, three conditions by cone index gradient [2][8]
Instrumentation	Smart-winch drawbar pull rig; OptiTrack, 32 cameras

Vehicles traverse prepared simulant and obstacles in this tank, and a drawbar pull rig can be positioned at the end of the bed to apply a chosen resistive force through a smart winch, so tractive performance is measured as a function of drawbar pull force [8]. Published dimensions differ between NASA documents and are not reconciled in any of them: the motion capture characterization gives 11.8 x 2.9 x 0.6 m, 11.8 x 3.0 x 0.3 m and 6.7 x 4.7 x 0.2 m for the three tanks [1]; the drawbar pull procedure describes a 12 x 6 x 0.3 m bin split into two equal lanes [2]. The operator’s 2025 figures are used above.

Source: [1]. Public domain (NASA / US government work).

GRC-1 is the defining material choice. Lunar simulants such as MLS-1, JSC-1 and JSC-1A reproduce composition, but they exist only in small quantities and would need vacuum and reduced gravity to exhibit lunar mechanical strength, which is impractical at vehicle scale. GRC-1 was therefore developed at Glenn as a strength simulant for Earth ambient conditions: a cohesionless sub-angular sand, D10 0.094 mm, D60 0.390 mm, specific gravity 2.583, whose cone index gradient can be conditioned to match the values measured by the Apollo 15 and 16 cone penetrometers [2][3]. The underlying assumption is stated explicitly, that if a dry granular terrain matches the measured lunar cone penetration observations then it will respond similarly to vehicle loading in compaction and shear resistance. The operator adds that GRC-1 targets the softer, more conservative end of equatorial lunar terrain strength with respect to vehicle mobility [8].

Three standard conditions are defined by target cone index gradient G. T1: G = 5.0 +/- 0.3 kPa/mm, relative density about 41 percent, bulk density about 1.71 g/cm3, seven passes of a 4.5 kg 8 x 8 in tamper dropped from about 8 cm [8]. T2: G = 3.5 +/- 0.5 kPa/mm, about 23 percent relative density, about 1.66 g/cm3, one winch-pulled pass of a 24 x 18 in lawn roller filled with GRC-1 to 172.5 kg; T3: G = 2.5 +/- 0.5 kPa/mm, about 11 percent relative density, about 1.63 g/cm3, loosening and leveling only. The relationship between cone index gradient and relative density for air-dry GRC-1 was established over more than 180 cone penetration tests and is considered reliable to 80 percent relative density, giving G = 0.0834 D_R + 1.5811 [8]. That relation is what makes the terrain a controlled variable rather than a nuisance parameter.

Preparation is manual and procedural, in three stages. Loosening returns the soil to its original state, without which compaction accumulates and density drifts upward over a test series. The published method inserts a flat-blade shovel perpendicular to the surface to its full depth of about 25 cm, pulls the handle back to about 45 degrees to lift and break up the soil, returns the shovel upright and lifts it out, then repeats 12 to 13 cm behind the resulting valley with about 25 percent overlap, over an area at least twice as wide as the wheel track [8]. Leveling first rakes out major hills and valleys, then draws a leveling blade that is suspended from above or carried on rails at the bin sides, so the blade never rests on the soil and cannot compact it. Compaction is either the tamper, which gives fine control through drop height and pass count, or the winch-pulled roller, which is faster and more consistent but offers little density control, since cone index gradient does not increase appreciably after the first pass. Verification is by cone penetrometer at eight locations, four along each wheel track centerline, before every vehicle test, judged on the gradient over 0 to 18 cm and on the linearity of the penetration-resistance curve, with target linearity coefficients of 0 +/- 0.015 kPa/mm2 for T1 and 0 +/- 0.01 for T2 and T3 [8].

Tilt bed

NASA/GRC/Bridget Caswell. Public domain (NASA / US government work).

Parameter	Value
Working volume	6 x 5 x 0.3 m tilting soil tank
Test article limits	The 400 kg Scarab rover has been run on it [3]
Simulant or terrain	GRC-1 as filled, changeable on request [8]
Slope	0 to 45 degrees, hydraulic lift; usable maximum set by the angle of repose
Instrumentation	Inclinometer on the bed; photogrammetric best-fit plane through the surface [1][3]

The tilt bed is the only slope capability in the laboratory, and it is set rather than measured: a hydraulic lift system takes the bed to 45 degrees of inclination, but the operator notes that in practice the maximum is limited by the angle of repose of the simulant in the bed rather than by the lift [8]. Slope angle was verified in earlier campaigns by an inclinometer on the side of the bed and, in later work, by fitting a plane through a photogrammetric point cloud of the actual soil surface [1][3]. Whether the bed can be moved under load is not published. Two cameras of the motion capture array sit at 4.9 m behind the tilt bed specifically so that the soil surface stays in view at maximum inclination [1].

Earlier documents describe the same bed as 7 x 5 x 0.3 m [2] and as approximately 6 x 4.5 m filled to 23 cm [3]; the operator’s 2025 figure is used above [8].

Sink Tank

Parameter	Value
Working volume	12 x 3 x 0.6 m, twice the depth of the other tanks
Simulant or terrain	Fillite, hollow alumino-silicate microspheres, 0.6 m bed
Instrumentation	OptiTrack, 32 cameras

Source: [8].

The Sink Tank exists to produce the failure case rather than the nominal one. Fillite is a low-density hollow microsphere medium with extremely low bearing and shear strength, chosen to mimic the wind-blown dune soil found on Mars that has produced entrapped conditions for rovers, of the kind that immobilized Spirit in iron(III) sulfate deposits [1]. The operator is explicit that it is not a high-fidelity simulant: it reproduces a trafficability case, not a material.

DRAGON M

Parameter	Value
Working volume	Dual-rail gantry over a soil bin, X 1850 mm, Y and Z 850 mm travel
Test article limits	Tooling rather than vehicles: auger, cone penetrometer, wheel-soil assembly
Simulant or terrain	GRC-3B
Instrumentation	Six-axis load cell, 1 kN Fy, 4 kN Fx and Fz, 250 Nm; string potentiometer

Source: [6].

DRAGON M is the automated successor to the manual three-stage preparation procedure: a rigid steel frame bolted to the floor around a soil bin, carrying a dual-rail gantry with a vertical drive mast and a rotating tool carriage on five servo axes [6]. X runs to 5 m/s over 1850 mm at 0.02 mm resolution and 1366 N; Y and Z to 1 m/s over 850 mm at 0.02 mm and 6146 N; A rotates continuously to 600 rpm at 0.02 degrees and 109 Nm; B continuously to 57 rpm at 0.02 degrees and 48 Nm. Tools include a 101.6 mm auger, a 60 degree cone penetrometer of 500 mm2 base area, and a wheel-soil assembly on the six-axis load cell, with a string potentiometer on the Z carriage reading sinkage [6].

It both loosens and compacts the full depth of a bin and characterizes it in place, which is what a ground contact model calibration needs: drag data and cone penetrometer measurements from a single preparation [6]. Loosening trials drove 96 auger holes distributed evenly across the bin with a 101.6 mm diameter, 67.73 mm pitch, 609.6 mm bladed auger at 400 degrees/s insertion and -360 degrees/s withdrawal at a 68 mm/s feed rate, followed by manual leveling with a wooden drag. Cone penetration at nine locations showed high consistency in tip resistance between 0 and 100 mm, increasing variability beyond 100 mm attributed to the limit of auger penetration or to re-consolidation at depth, and no measurable edge effects between central and peripheral locations [6]. DRAGON M does not appear on the operator’s current capability page and its present status is unconfirmed [8].

TREC

Parameter	Value
Working volume	Two adjacent soil tanks, each 3 x 2 x 0.75 m
Test article limits	Tires to 95 cm diameter; vertical load about 90 to 900 N
Simulant or terrain	GRC-3 in one tank, commercial #90 silica sand in the other
Gravity offload	Vertical tire load adjustable below vehicle weight, including offloading
Instrumentation	In-hub six-axis load cell, 900 N and 230 Nm; laser displacement sensor

Source: [8].

The TRaction and Excavation Capabilities rig is the laboratory’s single-wheel dynamometer. Carriage and tire are driven independently so that forced slip from 0 to 90 percent can be commanded, at linear speeds to 6 cm/s, and the in-hub load cell captures drive force and ground force together so a net traction versus slip curve comes out of a single run; a laser displacement sensor records tire sinkage against time [8]. GRC-3 is a custom silty-sand lunar simulant of semi-angular silica particles with Bonnie silt added for strength, aimed at in-situ resource utilization work such as excavation; #90 is a poorly graded silica sand representing aeolian Martian terrain. A semi-automated rake and screed method prepares both to repeatable conditions, and as in SLOPE the simulant can be swapped on request. Static and dynamic net traction over bedrock can also be measured on the rig.

Excavation Lab

Parameter	Value
Working volume	GRC-3B bin 76.2 x 183 x 76 cm; pea gravel bin 61 x 91 x 51 cm
Test article limits	Excavation tools on the arm end; 900 lbf (4005 N) vertical as installed
Temperature	Lab ambient, air conditioned
Simulant or terrain	GRC-3B, silica sand and silt matched to lunar particle size distribution
Instrumentation	156.5 mm six-axis load cell, 1300 N Fx and Fy, 3900 N Fz, 203 Nm

Source: [8].

The Advanced Planetary Excavator is a four degree of freedom backhoe-like arm with sealed rotary joints, 360 degrees of rotation about a vertical axis, over 2 m of reach and a 3 m maximum swing radius with bucket and load cell fitted [8]. It runs on electric linear actuators fed from facility 208 VAC three-phase power converted to 325 VDC at 100 A; joint speeds are 4.67 cm/s at the shoulder, 9.37 cm/s at the elbow and 11.71 cm/s at the wrist. Mechanical maximum vertical force is 1165 lbf (5184 N), controlled as installed to 900 lbf (4005 N) vertical and 400 lbf (1780 N) horizontal [8]. Load cell full-scale uncertainty is 1.25 percent in Fx, Fy, Tx and Ty and 2.00 percent in Fz and Tz.

Tools can be lowered to a chosen digging depth and driven at controlled rates along pre-programmed paths, under joystick control or planned trajectory for repeatability, and are either tethered to wall AC power or powered from the arm at a nominal 300 VDC [8]. The bin sits on an 81 x 188 x 36 cm shaker table of 22.24 kN capacity used to compact the simulant, the dump bin stands on a 61 x 61 cm platform scale reading to 2224 N in 0.22 N steps to weigh each delivered load, and a HEPA-filtered dust enclosure surrounds the arm and bins to protect operators from the respiratory hazard [8].

XCEL

Parameter	Value
Working volume	Environmental chamber 139 cm tall x 101 cm deep x 50 cm wide
Test article limits	Full-scale wheel or tire; maximum size set by fixture and stroke
Temperature	133 K to 370 K, liquid nitrogen or liquid carbon dioxide cooling
Instrumentation	Hydraulic load frame, nominal cyclic rate 1 Hz

Source: [8].

The Extreme Cyclic Environmental Load rig is a hydraulic load frame coupled to an environmental chamber sized to enclose a full-scale wheel, used for cyclic radial compression tests against a flat plate at planetary temperatures. The 133 K figure is the lowest tested to date rather than a floor [8].

XTERra

Parameter	Value
Test article limits	Tires to 95 cm diameter; vertical force to 1000 N, 2500 N peak
Temperature	Heated to 60 C; cooling possible, minimum temperature unknown
Simulant or terrain	Customizable regolith and rock terrain elements on a carousel
Instrumentation	Six-axis load cell at the tire; thermocouples; front and side cameras

Source: [8].

The Extreme Terrain Endurance Rig is a single-wheel life test rig. Carousel and tire are driven independently for forced slip from 0 to 100 percent, at linear speeds on the order of 100 cm/s depending on tire load, with continuous torque to 300 Nm and 500 Nm instantaneous [8]. Vertical tire motion is recorded through an inclinometer on the mounting arm [8]. It is where durability rather than traction is measured, over representative terrain elements including rocks rather than a prepared flat bed.

VOiD

Parameter	Value
Working volume	89 x 89 x 82.5 cm (35 x 35 x 32.5 in) internal
Test article limits	Sub-meter-cubed experiments; 4 ceiling mounts at 50 lb each
Vacuum	1e-6 torr, with filters and a throttle valve for pumpdown with simulant in
Temperature	-180 to +150 C, LN2-cooled shroud with internal heaters
Simulant or terrain	Regolith simulant in the chamber; product not specified

Source: [8].

The Vacuum Operations in Dust Environment chamber is the one rig in the group that supplies vacuum, and it is designed around the awkward part of that, pumping down with loose simulant present: a series of filters and a throttle valve allow a safe pumpdown with regolith in the chamber, and a thermal shroud cooled by liquid nitrogen dewars and heated by internal heaters sets the wall temperature. It is aimed at raising the technology readiness level of payloads past component-level testing. Customer feedthroughs run through fourteen DN-40 KF/QF ports, two DN-63 LF and five DN-250 LF; the DN-250 ports and two of the DN-40 ports are line of sight to the chamber center. The operator’s page states the rig was being built and was slated to be operational in summer 2024 [8].

Instrumentation

The motion capture array has grown since it was characterized. The 2022 report describes 16 Primex41 cameras on the laboratory walls: 14 at 4.3 m to clear the perimeter walkways and two at 4.9 m behind the tilt bed, so that the soil surface stays in view at maximum inclination [1]. Orientations were set at installation to maximize soil bin area in each field of view, with neighboring fields overlapping for redundancy, and four or more further cameras can be placed on tripods. The system tracks up to 300 rigid bodies in real time at a quoted plus or minus 0.1 mm and 0.5 degrees [1]. The operator’s capability page dated 31 March 2025 gives 32 cameras, 20 permanently wall-mounted and 12 on tripods, providing a full 360 degree field of view at sub-millimeter accuracy [8].

The accuracy figures were measured on the 16 camera configuration and bound the current array rather than describing it [1].

Source: [1]. Public domain (NASA / US government work).

The array was validated rather than assumed. An eight-LED OptiTrack active puck was mounted on a Thorlabs LTS300 300 mm precision linear stage under MATLAB control, elevated on aluminum extrusion and two heavy tripods to keep it out of the simulant, and stepped across its travel in 10 mm increments with about 600 OptiTrack samples taken at 100 Hz at each setpoint [1]. The bins were divided into 12 cells, the stage placed at three random locations in each, and the resulting error compared against the wand error Motive reports after calibration. Mean reported wand error was 0.17 mm against an independently calculated -0.06 +/- 0.65 mm over 2,249,327 measurements across the capture volume [1].

Bed state is measured rather than assumed. The laboratory standard is a Rimik CP40II cone penetrometer with a 30 degree apex and a 20.3 mm base diameter cone of 323 mm2 base area on a reduced-diameter shaft, pushed manually at a target 3 cm/s, recording average cone index over 10 mm depth increments to 18 cm [2].

Before the motion capture installation, vehicle tracking used stereo photogrammetry: coded two-dimensional targets on the wheels, a fixed camera pair triggering every 2 s, and GOM Pontos to recover three-dimensional target coordinates [3]. Terrain surface was recovered by projecting a speckled dot pattern onto the bed and processing the stereo pair in GOM Aramis, averaging each point over a 19 x 19 pixel square with centers 15 pixels apart, or 22.5 to 30 mm on the terrain, into a point cloud used as the sinkage datum. Camera resolution was 4272 x 2848 pixels for a spatial resolution of 1.5 to 2 mm, with calibration deviation between computed and actual coordinates averaging 0.196 pixels, roughly 0.29 to 0.39 mm [3]. For wheel-soil flow visualization the laboratory keeps a small bin with a clear glass wall against which a tire can be positioned and the soil displacement imaged through the sidewall [4][5]. The Shear Interface Imaging Analysis Tool, developed on this glass-walled bin, runs a half-width wheel at half payload against the glass at 8 frames per second and recovers soil flow magnitude, flow direction and the shear interface beneath the wheel by optical flow. It found the shear interface is set mainly by rim tangential stress rather than by drawbar load, since its extent barely changes from 5 to 20 percent slip while drawbar pull quadruples over that range, and that a walking or pushed wheel fails the soil in general shear rather than the forced rotational failure of a rolling wheel [9].

What it does not reproduce

Gravity. The laboratory operates at 1 g and has no offload rig; only TREC can relieve part of the wheel load, and it does so at the wheel rather than at the vehicle [8]. The historical workaround, still the one in use, is to build test vehicles light enough that wheel loads match what would be expected at one sixth Earth gravity, as was done for the Surveyor Lunar Roving Vehicle 6x6 and 4x4 test vehicles and for the Mobility Test Article GM-1 at one fifth of the Lunar Mobile Laboratory mass [2]. That reproduces the load but not the gravitational body force in the soil column.

Vacuum, and therefore composition-driven soil behavior, in the mobility bins. GRC-1 exists because controlling the atmospheric conditions needed for a compositional simulant to show lunar mechanical strength is impractical at this scale [2]. The trade is explicit: SLOPE reproduces terrain strength and gives up composition, particle shape beyond sub-angular, electrostatic behavior and cohesion. GRC-1 is cohesionless, against the up to 1 kPa of cohesion observed in lunar regolith [2]. VOiD supplies vacuum and dust, but in a sub-meter-cubed volume, so no vehicle-scale test is run under vacuum here [8].

The strongest lunar terrain. Over its full density range in Earth ambient conditions, GRC-1 covers four of the five lunar cone index gradient values in the National Space Science Data Center set. The one that cannot be reproduced is a high terrain strength, which the operators note does not pose a mobility challenge [2]. The facility is therefore biased conservative: it can be harder than the Moon but not easier.

Bed repeatability at scale. Because of the size of the bins the soil has to be processed in place. Removing and replacing it systematically for each test, which is the best way to guarantee a homogeneous bed, was judged unrealistic without an automated mechanical system, so the manual three-stage procedure exists as the alternative [2]. DRAGON M is the response to that limitation and its consistency falls off below 100 mm depth [6].

Motion capture accuracy is not uniform. These figures describe the 16 camera array as characterized in 2022, not the 32 camera array the operator lists in 2025 [1][8]. Mean error across the 777 m3 capture volume is submillimeter and matches the reported wand error, but measured error in some cells exceeds that mean, attributed to camera occlusion by the TREC test rig and the stairwell. The recommendation in the characterization report is that additional cameras be placed to raise accuracy in those regions [1]. Even-numbered cells, over the GRC-1 lane, more often showed greater absolute localized error than odd-numbered cells, and cell 10, on the tilt bed, showed the effect on every day of testing [1].

Slope surface is not flat. Even after hand raking, terrain height varies across the tilt bed enough that wheel position alone does not give true sinkage; the relative distance between wheel center and terrain has to be measured, which is why the photogrammetric terrain point cloud exists [3].

Test throughput. The angle of attack campaign notes that test preparation and post-processing are time-intensive enough that repeat tests were not conducted unless there was a known error [3]. Single-sample results are the norm rather than the exception.

Campaigns run there

Drawbar pull method development, published 2017. The standing drawbar pull procedure for off-road planetary vehicles was written at SLOPE, covering terrain preparation, the three GRC-1 conditions, hitch geometry and the metrics [2]. Findings that constrain the method: cone index gradient corresponds to bulk density for dry granular soils and can be mapped to other mechanical properties, and because the cone penetrometer flew on Apollo, lunar G values can be used directly to set up lunarlike conditions on Earth. The report also establishes that hitch height matters, with a worked case for a drawbar pull coefficient of 0.2 at a hitch height of 29 percent of the wheelbase [2].

Push-pull locomotion, 2014 to 2015. An inching gait, in which the vehicle alternately extends and retracts between wheel sets rather than rolling continuously, was tested in a soil bin of approximately 8 x 3 x 0.3 m of GRC-1 against a controlled drawbar pull load [4]. Inching generated about 37 percent of vehicle weight in drawbar pull with pneumatic tires against 27 percent when rolling, and about 33 percent against 25 percent with rigid tires [4]. An earlier account of the same work reports a pull coefficient rising to 0.28 when rolling with a maximum inching force about 30 to 40 percent higher [5]. Rolling resistance was measured separately by towing a chassis on free-running wheels and found to be about 5 percent of vehicle weight [4]. The mechanism, a more efficient soil shearing method, was identified through the glass-walled single-wheel bin [4][5].

Angle of attack on slopes, published 2018. The 400 kg Scarab rover on 71 cm rigid wheels was driven on the tilt bed at 16 combinations of four slope angles and four angles of attack, in GRC-1 prepared to the loose T3-like condition of G = 2.5 +/- 0.5 kPa/mm, relative density 11 percent, bulk density 1.63 g/cm3, friction angle 31.5 degrees and essentially zero cohesion [3]. Each run was a constant 0.13 rad/s wheel speed for 30 s, tracked photogrammetrically against a projected-pattern terrain point cloud. The trajectory always trended downslope of the intended path except at 90 degrees angle of attack; at 10 degrees angle of attack the vehicle made no uphill progress on slopes above 9 degrees, and on the 15 degree slope it made no progress at either 10 or 30 degrees [3]. Sinkage increased with angle of attack, rear wheel sinkage running consistently greater than front and reaching 8 cm, and wheel slip increased with both slope angle and angle of attack, front and rear differing by under 0.5 percent [3].

Motion capture characterization, 2022. Four data sessions validated the OptiTrack installation across the 12 cells at tilt bed inclinations of 0, 10 and 20 degrees [1]. Error magnitude showed no positive correlation with tilt bed angle. The conclusion validates the system for characterizing mobility and tractive performance, with the localized-occlusion caveat above [1].

Scale model vehicle dynamics, published 2017. A modular six or eight wheel-pod vehicle was built to evaluate whether scale models can simulate lunar vehicle dynamics [7]. Wheel pods came in at 0.73 kg each, 28 percent over the 0.57 kg mass target. Turning off one pair of tires reduced distance traveled by 22 percent in one configuration and 53 percent in another, and the eight-pod configuration performed worse than the six-pod configuration [7].

Ground contact model calibration, 2025. Terramechanics, discrete element and finite-element ground contact modeling approaches are being calibrated against SLOPE single-wheel data, with GRC-1 parameter sets drawn from the literature described as a first pass toward consistency between sources [6]. DRAGON M was built to supply the consistently prepared soil volumes that this calibration needs, capturing drag test data and cone penetrometer measurements from a single preparation [6].

VIPER ground truth, published 2025 [10]. VIPER carries a torque transducer in each of its four actuated suspension joints and a quadrature encoder pattern machined into the inboard wheel rim that its wheel-well cameras can read, so it measures wheel normal force and sinkage directly rather than inferring them, a combination no prior planetary rover flew. Those measurements were designed to be inverted through Bekker and classical bearing capacity relations for cohesion, internal friction angle, bearing strength and modulus of subgrade reaction at the lunar south pole, a region so far constrained only by Apollo equatorial data and Yutu-2. The published slope versus mean wheel slip curves come from the MGRU mobility surrogate driven uphill and downhill in loose GRC-1 simulant at 15 to 20 percent relative density on the tilt bed, the terrestrial proxy for the ten 10 m rail segments across 0 to 15 degrees of slope the mission itself planned to drive [10].