Colmena

LINX, Instituto de Ciencias Nucleares, UNAM. Used with attribution on the rights holder’s terms.

The problem

Colmena is a swarm of five micro-rovers built by the Laboratory of Space Instrumentation (LINX) at the Institute of Nuclear Sciences of the National Autonomous University of Mexico, flown as an integrated payload on Astrobotic’s Peregrine Mission One [8]. The project’s stated end is mineral prospection: lunar surface resources are scattered over large areas, so the work is to be spread across many small, simple units that survey individually and act collectively [13]. Median regolith depth is 2.5 to 6.1 m across mare regions and 6.7 to 8.7 m across highland regions by crater morphology [4], and radar inversion gives about 5 m for mare and more than 10 m for highland terrain [6].

A rover of this class has to survive and work in conditions set by the surface and by a gram-scale mass budget. The regolith surface is at under 8e-9 torr, and dust adheres to metals at 0.01 to 0.1 N/cm2 [7]. The top 15 cm has a bulk density of 1.50 g/cm3 [5], with cohesion of 0.1 to 1.0 kN/m2 controlled by porosity [10]. Sub-micron grains are charged by ultraviolet photons; photoelectric yield measured on single returned grains falls by about an order of magnitude below roughly 1 micrometer radius and is asymptotic above a few micrometers [1]. The science case places the rovers inside the lofted charged layer [8]. At under 5 g of compressive force per wheel, the ground may behave as a fluid-like medium rather than the locked solid a heavy rover compresses [8]. At 608 g for the whole payload, no conventional radiation shielding was affordable, and the design had to survive standby from -180 to 150 C and operation from 0 to 120 C [8].

For scale, the Apollo lunar roving vehicle was 210 kg with 490 kg of payload capacity and covered 28 km at Hadley Rille on Apollo 15 [35]. A 1992 projection of rovers of about 5 kg built from micromachined sensors and VLSI computers had no hardware behind it [9].

What has flown

Colmena-1, January 2024. Peregrine launched on 8 January 2024 on a Vulcan [2]. A helium pressure control valve lost seal capability, most likely through vibration-loosened threads in a joint internal to the valve, and let helium flow into the oxidizer tank until it ruptured about 92 minutes after separation, ruling out a landing [2]. The lander operated for ten days and fourteen hours in cislunar space before a controlled reentry over the South Pacific on 18 January [2]. Colmena was never deployed. Its payload was turned on in cruise, which it was not designed for, and every operational mode, telecommunications, thermal telemetry and single event resilience were exercised [8]. The project reports 75 percent of technical objectives met with no per-objective breakdown, so every objective that needed contact with the surface is among the missing quarter [8]. No temperatures, dose figures or upset counts are published [8].

Iris, January 2024. Carnegie Mellon’s Iris rover shared Peregrine and, with the landing ruled out, was never placed on a surface [2].

SLIM, LEV-1 and LEV-2, January 2024. SLIM landed on 19 January 2024; fourteen vision-based navigation fixes were taken down to the vertical descent, and the first 50 m hover image placed the lander 3.4 m from the planned target against a 100 m requirement [23]. Japan’s 2.1 kg LEV-1 was deployed just before touchdown and completed seven autonomous hops, received on a direct-to-Earth S-band link at Usuda and Uchinoura [21]. It ran about two hours on the surface [24]. LEV-2, a roughly 8 cm sphere, righted itself, drove away from the lander, selected images onboard and transmitted them, with a confirmed minimum of about 108 minutes of operation [22]. Two image data sets from LEV-2 reached Earth through LEV-1, which exercised a rover-to-rover link on the surface [21]. The displacement between LEV-2’s two returned images was about 0.13 m, derived photogrammetrically from one rock [22]. LEV-1’s single wheel carries a 90 degree tilted octagonal grouser chosen with a resistive force theory model; the draw-bar-pull and sinkage numbers behind that choice are Earth-gravity tests in terrestrial silica sand [24].

Colmena-1 has no surface result of its own. The surface evidence for micro-rover operation, and for one rover relaying through another, is the LEV pair.

What is being tried now

Deployment by catapult (ground tested, flight payload built). The rovers ride in a bucket under a removable shroud on the arms of a catapult in the deployment module, which stays bolted to the lander deck. The catapult is armed during integration and released by a cutting system after separate arm and trigger commands from the ground [8]. It throws the rovers 10 to 15 m, clear of the lander’s shadow and of regolith disturbed by the descent thrusters, and it randomizes their distribution [8]. The disturbed-regolith hazard is documented from Apollo 12: Surveyor 3, 155 m from the Lunar Module, carried 103.1 pits per cm2 on a returned camera shroud coupon [14]. The price of random placement is that landing attitude is unknown, so each rover is reversible about its mid-plane, with two wheels, four skates and solar panels on both faces [8]. The catapult has not fired on the Moon.

Dusty plasma science from a swarm (proposal, partly simulated). The rovers carry magnetic field sensors from 1 Hz to 100 kHz, accelerometers, magnetometers, temperature sensors and an electromagnetic plasma probing subsystem, and exchange signals among themselves so that attenuation measures the charged layer [8]. Dusty-plasma coupling to the rovers is simulated, not measured [8]. Lofting is supported by laboratory work: insulating grains hop at about 0.6 m/s under ultraviolet or plasma exposure, equivalent to 0.11 m in lunar gravity [12]. A theoretical model puts the near-surface levitated dust density near 1e3 per cm3 [11].

Micro-rover autonomy (ground tested and simulated). For the MoonRanger micro-rover, the lack of a radio link to Earth outside lander Wi-Fi range forced onboard autonomy, and dead reckoning from IMU and wheel encoders alone drifted about 5 percent in field data at an analog site [17]. MoonRanger’s thermal management is described as four behaviors: 13 thermistors, steering to trade sun heating against power, computer power throttling and resistive heaters, with no thresholds published [16]. Coverage path planning for lunar micro-rovers has been simulated on an Apollo 14 elevation model and field tested with the 7 kg CLOVER rover, reaching 80 percent coverage with a 0.41 m mean localization error on a 5 by 5 m grid [18]. No navigation method for Colmena’s rovers is published, and none was exercised [8].

Swarm coordination (ground testbeds and simulation). A Markov-chain guidance law has been simulated with 1000 notional agents and demonstrated physically only with a handful of ground robots on a 9-cell grid [3]. NASA Ames’ Troupe testbed runs 4 to 10 ground rovers for autonomous mapping, on commercial radio-control chassis with no mapping accuracy reported [19]. Spacebit’s Asagumo was manifested on Peregrine [20]. No swarm of surface robots has operated cooperatively on the Moon.

Thermal design at micro-rover scale (ground modeled). The Canadian Lunar Rover Mission, as originally proposed, set a 30 kg rover and a 1 hour operating target inside shadowed regions, a design target since superseded [25]. Colmena’s 0 to 120 C range is a requirement, not a result [8].

Dust mitigation (laboratory, one ISS flight). Dust degrades thermal control surfaces sharply: solar absorptivity may rise by as much as 50 percent at 12 percent area coverage [31]. Brushing restored white paint absorptance to emittance ratio from 0.65 dusted to 0.30 after 20 strokes in a bell jar with simulant, and bench brushing results did not transfer to that environment [15]. Passive protection on VIPER is a 42-layer outer barrier over each mobility module with labyrinth, felt and spring-energized PTFE seals, tested at ambient pressure [29]. A weighted trade study ranks a spring-loaded Teflon shaft seal first at 21.5 against 19.5 for the next, a subjective score and not a leak rate [32]. Active removal: electrodynamic dust shields recovered at least 90 percent of solar panel output within two minutes in vacuum and cleared particles up to 450 micrometers in parabolic flights at lunar gravity [26]. Twelve panels flew on MISSE-11 on the ISS, two energized at 4 kV and 10 Hz, and no dust removal was measured on orbit; the panels showed current spikes of unknown cause [27]. An electrostatic traveling-wave cleaner reached 90 percent cleaning of simulant at 800 V in vacuum, 98 percent with ultrasonic vibration, and was also tested on Apollo 11 soil [30]. For wear, a 4 minute exposure to simulant at 105 m/s cut visible transmittance of specimens from 86 to 29 percent [33], and a single 10 mg shot at about 360 m/s roughened surfaces by a factor of 8 [34]. A survey of abrasion work found most tests run at room conditions rather than in vacuum and at temperature [28]. No dust-removal or sealing device is published for the 56 g Colmena rover [8].

Colmena follow-on missions (plans). Colmena-2, planned for 2028 [8], is to deploy at least three prospection rover prototypes with a positioning system, beyond-the-horizon navigation and one night of hibernation. Colmena-3, planned for 2031, adds sample collection and transport and survival across several diurnal cycles [8].

The numbers

Parameter	Value
Rovers	5, identical
Rover mass	about 56 g each, 280 g total
Rover envelope	approximately 12 x 12 x 4 cm, torus about 1 cm thick
Wheels	2, plus four skates
Deployment module mass	328 g
Payload total mass	608 g
Standby temperature range	-180 to 150 C
Operating temperature range	0 to 120 C
Radiation survival requirement	120 days in cislunar space outside the magnetopause
Catapult range	10 to 15 m
Compressive force per wheel	under 5 g
Launch	8 January 2024, Vulcan [2]
Time to tank rupture after separation	about 92 minutes [2]
Duration operated	10 days 14 hours [2]
Technical objectives met	75 percent, project accounting

Rows without a marker are from [8]; the launch and anomaly rows are from [2]. Speed, drive torque, battery capacity, generation, processor and radio parameters are not published [8].

Small lunar rover, January 2024	Outcome	Source
Colmena-1 (five rovers, 56 g each)	Powered in cruise, never deployed	[8]
Iris	Aboard Peregrine, never placed on a surface	[2]
LEV-1 (2.1 kg)	Seven hops, about two hours on the surface	[21][24]
LEV-2 (about 8 cm sphere)	Imaged from the surface, relayed through LEV-1, 108 minutes minimum	[21][22]

What is not established

No Colmena rover reached the surface, so deployment by catapult, landing orientation, traction at under 5 g per wheel, the fluid-like upper dust layer and every plasma measurement remain unmet and untested [8]. The 75 percent figure is the project’s own accounting with no per-objective breakdown [8]. The cruise operation was improvised, and no temperatures, dose figures or upset counts are published from it [8]. The surface layer scale quoted for the target is cited context rather than a result [8]. The root cause of the lander loss is stated as most likely, the flight valve was never recovered, and the report is the operator’s own [2]. The photoelectric yield, lofting and dusty-plasma literature behind the science case comes from laboratory work on returned grains or simulants and from theory [1][11][12], and no in-place measurement of the layer exists. The dust mitigation results above are ground tests with simulants, and the one flight, MISSE-11, was in low Earth orbit, not on the Moon [27]. The LEV-2 displacement figure rests on two images [22]. Colmena-2 and Colmena-3 dates are the project’s plans [8]. The project page is the team’s own and is not independently verified [13].

References