Testing

Verification methods, facilities, and simulants used to qualify robotic hardware for environments that cannot be reproduced in full on Earth. Each page covers what the method reproduces, what it does not, the fidelity limits that have caused flight surprises, and the facilities where the work is done.

What ground testing has to do

A planetary robot is qualified on the ground against vacuum, temperature extremes, reduced gravity, radiation, abrasive and cohesive dust, and a soil whose mechanical properties are known only from a few sites. No single rig reproduces more than a subset. Combined lunar-environment testing is a design problem in its own right, because the environments constrain each other: dusty vacuum operation changes how a chamber is cleaned, pumped and accelerated [7]. The test program for a flight rover is therefore a set of partial simulations, each chosen for the failure it can expose, and the engineering question is which residual risk each leaves behind.

Families of method

Thermal vacuum and solar simulation. The oldest family. The measured spectral mismatch of the JPL 25-ft space simulator showed that an absorptance computed against the solar spectrum is wrong for any coating with a spectrally selective response [6]. Mars 2020 system thermal vacuum and thermal model correlation are covered on thermal vacuum testing.

Dust and abrasion. Apollo is the baseline: the catalog of dust effects on extravehicular systems traces to primary mission documents [11]. Ground work since then splits by what is measured. Three-body abrasion of common spacecraft materials against lunar simulants gives wear volumes under one controlled dry configuration [12], and abrasion of stainless steel rotating parts against Moon and Mars simulants joins simulant characterization to mechanism life [13]. Abrasion at lunar temperature extremes, which room-ambient tests do not cover, has been measured on Ti-hBN coatings [14]. Coating down-selection for dust tolerance has covered adhesion and thermal shock as well as wear [15]. The dust sealing architecture of a flight lunar rover is published with materials, compressions and labyrinth gaps [16]. Dust adhesion is the case where ambient and vacuum disagree: brushing worked in air and failed in a simulated lunar environment, which matches the Apollo outcome [10], and the requirements for the bell jar that exposed this are documented [9]. Triboelectric charging of candidate insulators has been measured on the insulator rather than the grain [43]. Details are on dust testing.

Regolith simulants. A simulant is a chosen subset of properties of a real soil. Its producer lists what it does not simulate [20]. A selection guide scores multiple lunar simulants on a common basis [17]. A simulant reformulated for operator safety rather than fidelity has published exposure measurements and a cone index gradient [18], and size-fraction-resolved geotechnical data separates flow behavior by grain size [19]. The reference values come from Apollo and Surveyor measurements [23][24], and returned Chang’e-6 farside soil is the one held measurement on real returned material that shows strongly cohesive behavior [22]. Asteroid simulants are at a proposal stage, organized by spectral class [21]. See regolith simulants.

Radiation and electrical environment. Single event testing follows a practice that originates in a heavy-ion guideline: delid the part, hold vacuum, measure flux and uniformity independently of the accelerator, and tilt the part to vary effective LET [25]. For total dose the early hardness-assurance survey found no pre-irradiation electrical screen correlating with radiation response [26]. Electromagnetic compatibility on Mars 2020 used a priority rubric to cut the source-victim pair list and measured the cost of removing shields from twisted-pair drive lines [28]. See radiation testing.

Vibration and subsystem verification. The fatigue-equivalence trade between level and duration and the tolerance bands on random vibration are laid out in a 1971 text that later standards state without justification [27]. A landing radar was tested with a simulator that generates the return per pulse from a terrain model, plus delay-line end-to-end checks [29]. Verification of reused robotic hardware, the practical problem behind heritage claims, is documented for the SPDM [33].

Mobility, sampling and field analog. Best practices for planetary roving vehicle mobility testing connect each lunar terrain vehicle requirement to a test method and a named gap [31]. The Mars 2020 sampling qualification campaign recorded cores, abrasions, seals and bits consumed [30]. A simulated search-for-life mission with automated drilling quantified operator command cycles [42]. Reduced gravity is covered on reduced gravity testing and facility envelopes on the facilities pages; a lunar test bed of basaltic cinders over coarser material is described only by a promotional web page [41].

Simulation and analog media. Underwater testing of space robotics is validated by time scaling the plan so both platforms share a disturbance robustness degree [32]. A framework of key performance indicators for multi-robot lunar mapping defines efficiency, workload and autonomy measures for field trials [34]. See simulation and digital twins.

What has been checked against flight

Ground qualification is checked against the flight it was predicting only rarely, and the matched cases are specific rather than general. Curiosity’s accumulated flight thermal cycles have been rainflow counted against the accelerated cycling test run on the same board design [1]. The MSL generator’s output was measured against its pre-flight prediction at three points before launch and once on the surface [2]. Mast pointing accuracy achieved on Mars was carried into the Mars 2020 requirement as an operational figure with outliers rather than as the ground expectation it replaced [3].

Other comparisons are of a different kind: a design lifetime against what flew. The Mars Exploration Rover 90 sol mission expectation was wrong for solar array dust because accumulation was slower and cleaning more frequent than predicted [36]. Spirit’s end of mission report gives whole-of-mission actuator revolution counts against tested life [35], Opportunity’s loss to a dust storm is documented with the tau and energy trend of its final sols [37], Ingenuity operated outside its design envelope through a Martian winter [38], and InSight arm activities tie commanded actions to measured outcomes with testbed force calibration [39]. Terrestrial dust-emission climatology is the contextual reference for how dust events are forced and timed [40].

Facility capability parameters have almost no counterpart of that kind. Chamber base pressure, bed density and preparation state, slope, illumination and gravity offload fidelity are published as what a rig delivers to an article standing on or inside it, and none of them is re-measured on the hardware once it is working in the place the rig stood in for. A facility envelope is therefore a capability statement and not an error bar against flight, and most stated fidelity limits are arguments from physics rather than measured discrepancies. The one lunar analog facility held with a full geotechnical, abrasivity, mineralogical and chemical characterization of the material in its beds is LUNA [8].

Two thermal vacuum cases are the exception, and both are on flight vehicles. The Mars 2020 system thermal vacuum test concluded that a cruise thruster catalyst bed heater could stay on continuously from launch, which flight contradicted, because the chamber simulated no off-Sun angle flux [4]. Perseverance’s correlated thermal model inherited the 4 percent voltage loss of the ground support equipment cabling in place of the 9 percent of the flight flex cables, which produced preheat ramp rate errors above 5 C in flight until the cable loss was corrected [5]. Both are covered on thermal vacuum testing.

What is not established

No held source measures the offset between a chamber or bed result and the same quantity on the surface for the same hardware, so every facility fidelity statement above rests on physics arguments. Dust adhesion results from ambient air are known to mislead [9][10], but the size of the error in vacuum is reported only for the cases tested. Heritage verification of reused hardware is documented for one program [33]. Asteroid simulants are proposed and not yet validated against returned samples [21]. Lunar farside soil properties come from one returned sample set [22]. Mission-level comparisons of qualification life against flown life exist for a few vehicles [35][36][37] and do not cover the manipulator, drill or mast classes systematically.