Micro-Nova

The problem

The lunar poles hold water ice inside permanently shadowed regions and on slopes and blocky ground that a wheeled rover cannot enter. Reaching that terrain calls for a vehicle that does not depend on wheels, survives landing on its own propulsion and operates where sunlight is absent or grazing. Craters and pits add a descent-and-return requirement: the vehicle must go down a wall no rover could drive and come back out. The engineering record for ground properties is Apollo-era, with nothing in it measuring polar regolith [34].

Micro-Nova, also called uNova and flown as S. P. Hopper, is Intuitive Machines’ answer: a rocket-propelled hopper with its own propulsion, avionics, power, flight control and communications, deployed from a lander, that flies point to point and lands under a precision landing and hazard avoidance system that images the surface and guides the vehicle itself [1]. The vehicle is a single structure sized between 30 and 60 kg by its propellant tank, trading one-way range against payload mass [1]. Intuitive Machines lists hopper services beside fixed surface delivery, rover delivery and satellite delivery as one of four service lines [2].

What has flown

Micro-Nova on IM-2 (March 2025). The first unit was a secondary payload on the Nova-C lander Athena, with NASA’s PRIME-1 ice-prospecting suite and a Nokia 4G/LTE surface communications demonstration on board [5]. Its three technology objectives, in order, were deployment from the host lander, independent power generation and management on the surface, and end-to-end communication between hopper and lander [5]. Athena touched down on 6 March 2025 about 250 m from its intended site, on its side inside a crater at Mons Mouton [8]. With the solar panels misoriented and the crater extremely cold, the company did not expect the lander to recharge and declared the mission over the next day [8]. The hopper was never released and performed no hop.

LEV-1 on SLIM (January 2024). The only hopper to operate on the Moon is mechanical and small [13]. LEV-1 (2.1 kg) separated from the SLIM lander during the landing sequence on 20 January 2024, made six hops confirmed by the change in sensed gravity across each, and chose hop direction onboard from camera imagery. It transmitted directly to Earth for 1 hour 51 minutes before battery depletion [13]. The flight record gives no hop distance or height, so its locomotion performance is unquantified [14].

Lander-hosted precision landing on SLIM. SLIM took fourteen navigation fixes from orbit through vertical descent; comparison of the 50 m hover image with the target gave a 3.4 m position error against a 100 m requirement [15]. One of its two main engines then lost most of its thrust and the lander came to rest tilted [16]. The flight is one sample, so the error is an outcome, not a distribution [15].

Instruments on other lander-hosted flights. LISTER on Blue Ghost Mission One reached 0.98 m against a 2 m threshold, and its first heat flow value is preliminary [17]. Lunar PlanetVac collected 6.5 cc in one sampling event, from a head position the team could not reconstruct [18]. PRIME-1, on the same lander as the hopper, paired the TRIDENT drill with the MSolo mass spectrometer; no drilling into lunar regolith was performed, and only mechanism checkouts inside regolith thrown up at landing were exercised [12]. Nine lunar missions launched from 2019 to March 2024 and four ended in failure [3].

What is being tried now

Propulsive hoppers. Micro-Nova is the only propulsive lunar hopper built for flight, and it is at zero flown hops [8]. The extended configuration is specified at over 30 km one way, or a 4 km deep crater descent and return, both trajectory calculations [1]. A later company overview quotes 8 kg of payload over more than 25 km and a height of about 1 m, which does not match the 5 kg and 70 cm of the earlier specification; neither is a flight measurement [2].

Rotorcraft and propulsive drones. A lunar reconnaissance drone with a towed service station was sized to map at 0.1 m per pixel with flash lidar, at 9 km cumulative range over 11 flights without refueling [9]. It is a pre-hardware design study with no prototype built or tested [9].

Legged and jumping robots. Actuator-driven jumpers carry no propellant, so a tank does not limit their hop count. None has flown. A 1.25 kg underactuated jumper has jumped on a laboratory bench in Earth gravity, reaching 0.59 m high and 0.82 m forward [22]. A 12.5 kg quadruped learned jumping on a cable-offload rig whose effective gravity varied with a tension fluctuation of 0 to 1.2 kg equivalent [21]. A quadruped attitude policy reached hardware on an air-bearing stand, one rotational axis at a time, while walking and jumping results stay in simulation [30]. SpaceBok, 22 kg, learned cat-like jumping and landing, with hardware tests only in two dimensions [31]. A 12.45 kg tethered 3-DOF hopping leg is a terrestrial laboratory platform [32]. Two tethered robots that control attitude by inertial morphing are a concept studied only in a simulator [23], and an 8.2 kg reaction-wheel bipedal hopper has been modeled at lunar gravity on synthetic cratered terrain, with 0.8 m per hop [20]. LunarLeaper, a legged robot for lunar pits, is at pre-Phase-A with early estimates of 32 kg and 164 W peak power [24]. Pit access by tether is the Moon Diver approach, an Axel rover that rappelled slopes at 10 cm per second in Earth field tests [33].

Small impact-tolerant rovers. Dymon’s YAOKI is offered as a rover that withstands impacts of up to 100 G so that it can be thrown into lunar caves, a claim with no stated test standard [25].

Navigation and hazard detection for landing and hopping. The SPLICE hazard detection flight software, tested on post-processed lidar elevation maps from a built hazard field, reached true positive rates of 50 to 70 percent and false positive rates of 0.4 to 0.6 percent at operational downsampling [19]. Terrain-relative navigation depends on orbital maps; the LOLA 5 m per pixel elevation map near the south pole carries data in only about 10 percent of its pixels, the rest interpolated [29]. A decoupled quaternion attitude law for Chandrayaan-3 landing has been evaluated only in the authors’ simulator [28]. Moon Diver’s terrain-relative navigation ellipse of 100 m is a design value, not demonstrated on the Moon [33].

Terrain knowledge for polar sites. A 1 m per pixel elevation model underlies a water-ice prospectivity model for the Mons Mouton VIPER area, built from orbital data with weights its authors call plausible but arbitrary; it contains no surface measurement of ice [26]. Orbital estimates of median regolith thickness for four candidate Emirates Lunar Mission ellipses range from 1.2 to 2.9 m, again without surface verification [27]. An open 0.30 m elevation model exists for the Chandrayaan-3 Vikram site, with a median triangulation error of 8.1 cm [11], [10]; none is published for the Athena site.

The numbers

Parameter	Value	Source
Height	70 cm	[1]
System mass	30 kg, demonstration configuration	[1]
Fueled mass, S. P. Hopper	29 kg, with 6 kg further propellant capacity	[5]
System mass, extended	60 kg, same structure with a larger propellant tank	[1]
Payload capacity, extended	5 kg	[1]
Maximum one-way range, extended	over 30 km	[1]
Crater descent and return, extended	4 km deep	[1]
Company overview figures	8 kg payload, more than 25 km, about 1 m tall	[2]
Guidance	precision landing and hazard avoidance from surface imaging	[1]
Instruments	medium angle camera (39 by 51 degrees), 186 degree horizon camera, navigation camera, lunar radiometer	[5]
Radiometer accuracy	10 K at 70 K, 5 K at 100 K	[5]

The range and crater figures are trajectory calculations for the larger configuration, not flown results [1].

Mission item	Value	Source
Landing	6 March 2025, Mons Mouton	[8]
Lander attitude	on its side, inside a crater	[8]
Landing offset from target	about 250 m	[8]
Hops performed	none	[8]
Planned deployment hop	about 20 m by rail launch	[5]
Planned flight 2	90 m, propellant-optimized, with hazard avoidance	[5]
Planned flight 3	300 m at a constant 100 m altitude	[5]
Planned stereo imaging	3 cm per pixel over more than 20,000 square meters	[5]

The abstracts written before flight describe a landing on the Spudis crater side of the Shackleton to de Gerlache connecting ridge in late 2022 or early 2023 [5]; the mission that flew landed at Mons Mouton.

Companion payload	Value	Source
TRIDENT drill mass	about 22 kg, plus about 7 kg avionics and harness	[7]
TRIDENT design depth	0.91 m (about three feet); specification sheet 1020 mm	[6], [7]
TRIDENT auger	120 rpm, 5.5 N.m average torque, 87 W average power	[7]

The Nokia LTE system aboard Athena was intended to demonstrate lander-to-rover links at 200 to 300 m and up to 2 km [4].

What is not established

No hop of any length has been performed by this vehicle. Every propulsion, guidance, power, thermal and communications figure published for it is a pre-flight design value or a trajectory calculation for the 60 kg extended configuration, not a measured or flown result [5], [1]. Propellant mass, specific impulse, hop accuracy, landing dispersion, power budget, thermal limits, processor, and the band, data rate and range of the hopper-to-lander link are not published for either configuration [1].

The demonstration existed to prove deployment, independent power management and hopper-to-lander communication; none was reached because the hopper was never released from a lander resting on its side [8]. No public account gives a cause for the landing attitude, descent telemetry, touchdown velocity or slope [8]. The Nokia paper predates IM-2 and records the network as an objective, not a result [4].

Hazard avoidance and terrain relative navigation were named among the technologies the demonstration was to prove, so neither was flight proven on this vehicle [5]. SPLICE, SLIM and the terrain maps were tested on other platforms or offline, and their performance on a hopper is untested. Every legged and jumping robot above is a simulation, bench or cable-rig result, not a lunar one. Whether a second Micro-Nova will fly, and in which configuration, is not stated in the cited sources.