Asagumo

Overview

A lunar lava tube offers what a surface mission cannot: shielding from radiation and temperature swing, and a plausible cache of ancient volatiles. Reaching its interior means descending a skylight or crawling past a rockfall into terrain no wheeled vehicle can cross, with no line of sight to a lander for either communication or light. Asagumo was Spacebit’s answer to the first half of that problem: a four-legged, single-unit CubeSat-frame walking micro-rover, developed by the British company as a technology demonstration for a lava-tube exploration program, and manifested as a payload on Astrobotic’s Peregrine lander [1]. It was not aboard when Peregrine Mission One launched in January 2024 [2], and Spacebit’s own published material describing it is no longer served. It was one of a number of small technology payloads that a JPL survey of commercial lunar lander and rover missions groups alongside Asagumo in the early 2020s, most carrying a single demonstration objective rather than a science payload [3].

Walking rather than rolling is the design space’s answer to unstructured, broken terrain that a wheel cannot span: legged locomotion research since the earliest work on quadruped creeping gaits has treated a leg as a mechanism for placing discrete footholds and rejecting disturbance, not for rolling contact, and the dynamically stable running and hopping machines that followed kept that same premise [4][5]. JPL’s own limbed platforms follow the same reasoning at opposite ends of the design space: ATHLETE trades wheels for a hex-limbed, wheel-tipped mobility system to place a foot precisely during outpost assembly and regolith construction tasks a rolling chassis cannot do [6], and a microspine-gripping climbing robot built for cliff faces and lava tube walls trades rolling traction entirely for an anchor that holds in any orientation on natural rock [7]. Legs and grippers cost more mechanism and control complexity than wheels, and both are chosen only where the terrain or the destination requires a foothold or a grip that a wheel cannot make. Asagumo’s own literature states that reasoning explicitly: the destination, not the surface texture at the landing site, is what motivated the leg architecture, since a swarm of walking rovers was to be carried to a site by a wheeled vehicle and deployed into lava tubes that a wheeled vehicle could not enter.

The demonstration itself was modest and specific: deploy from the lander, walk at least 10 meters from it under teleoperation through the lander’s wireless network, and validate the legs, wide-field cameras and three-dimensional lidar over up to eight days on the surface [1]. It never reached the Moon to attempt that. Peregrine’s own post-mission report attributes the loss of the lander to a propulsion valve failure roughly 92 minutes after separation, unrelated to any individual payload, and lists legs and terrain-relative navigation among the lander subsystems that were never exercised at flight readiness because the mission did not reach the surface [2].

Specifications

Parameter	Value	Source
Mass	1.3 kg	[1]
Envelope	single-unit CubeSat frame, 10 cm on a side
Legs	4
Power	solar
Sensors	wide-field cameras and three-dimensional lidar
Planned traverse	at least 10 m from the lander
Planned surface duration	up to 8 days
Control	teleoperation over the lander’s wireless network

The published record for this vehicle is thin. It consists of a conference abstract, whose full paper is not openly available, and company material that is no longer online. No mass breakdown, actuator count, gait, walking speed, step height, power figure, battery capacity, thermal limit, processor or radio parameter appears in any source obtainable here, and no peer-reviewed publication exists. That gap is not unusual for a legged planetary vehicle at this stage: even mature quadruped programs aimed at the Moon and Mars still report gait, sinkage and attitude-control performance from reinforcement-learning simulation and single-robot test rigs rather than flight hardware [11][12].

Mission profile

The vehicle was manifested on Peregrine and did not fly on it [2]. No other delivery has been announced.

Mobility

Four legs on a 10 cm frame, at 1.3 kg [1]. Neither the joint count per leg, the gait, the walking speed, the step height nor any measured mobility result is published.

The reason for choosing legs over wheels is stated as the destination rather than the terrain: the program’s target is the interior of a lava tube, reached by deploying a swarm of walking rovers from a wheeled carrier, and the surface demonstration on Peregrine was to validate the leg system before that [1]. Contemporary lava-tube concepts converge on the same answer for the same reason. A limbed robot that locomotes by extending and retracting its limbs to grip a tunnel wall, and a combined legged-and-wheeled rover built specifically for a Martian or lunar lava tube interior, are both designed around foothold placement and climbing rather than rolling traction, because the terrain past the entrance is exactly what a wheel cannot cross [9][10]. What none of these programs, Asagumo included, has yet published for a lunar-relevant surface is a measured foot-terrain interaction: laboratory work on a legged gripper’s sinkage in granular material shows sinkage depth increasing with slope angle as the surface collapses more readily underfoot, a regime any of these vehicles would meet on a lava tube’s rubble slope [8].

Power and energy

Solar [1]. No generation, storage or duty cycle figure is published, and how a 10 cm vehicle with a fixed array would work in the near-total darkness of a lava tube interior is not addressed in any available source.

Thermal

Not published. Protecting the robots thermally during the transit to the Moon, and building a lightweight deployment mechanism that does so, is named as one of the program’s major design challenges [1].

Compute and avionics

Not published. A comparable lunar-regolith quadruped concept runs its locomotion policy onboard, trained by reinforcement learning against a physical model of the granular surface-foot contact rather than assumed rigid contact, because a policy trained on rigid ground destabilizes on soft regolith [11]; no equivalent architecture has been described for Asagumo.

Autonomy

Teleoperation over the lander’s wireless resources was the flight baseline, and autonomy was the development problem rather than the flight capability [1]. The stated approach is specific: executing a mission through communication disruptions using proprioceptive sensors and position data for terrain navigation, rather than depending on camera and lidar input. That choice follows from the destination, since a robot inside a lava tube has neither a reliable link to a surface relay nor useful ambient light for a camera. It is the same constraint that has since pushed subterranean multi-robot programs toward modeling and predicting radio signal strength from the local geometry, and toward coordination schemes built to tolerate and route around intermittent connectivity, in place of the assumption of a continuous link that holds on an open, sunlit surface [14][15].

Communications

Through the lander’s wireless network. Band, rate and range are not published. Comparable subterranean multi-robot programs treat the link itself as something to be modeled and managed rather than assumed: predicting received signal strength from the surrounding geometry, and routing data around nodes that drop out, are both built to cope with the same class of intermittent, non-line-of-sight connectivity a lava tube presents [14][15].

Payload and instruments

Wide-field cameras, one of which was to image the rover itself, with full high-definition video capability, and a three-dimensional lidar [1]. On the Peregrine demonstration these were the subjects of the validation as much as its instruments.

Modes of operation

Not published. Asagumo’s cancellation before flight puts it among the commercial lunar payloads of the early 2020s that were announced for a manifest but never demonstrated on the surface [3].

Ground operations

Not published beyond teleoperation through the lander. Field programs aimed at the same destination have instead run their operations concepts at terrestrial lava tubes, testing EVA suit and equipment requirements against real rubble, darkness and confined passages before committing to flight hardware [13]; no equivalent field test of Asagumo has been published.

Technologies developed

Nothing was flown and no measurement has been published. What the program contributed to the record is a stated design position: that a lava tube mission is a navigation problem before it is a mobility one, and that the sensing which works there is proprioceptive rather than optical, because neither the communication link nor the illumination that cameras and lidar depend on can be assumed underground [1]. That position is not unique to Asagumo: the limbed climbing and hybrid legged-wheeled concepts aimed at the same destination converge on mechanical grip and onboard perception rather than a surface robot’s usual reliance on a relay and ambient light [9][10].

What is not established

No mass breakdown, actuator count, gait, walking speed, step height, power budget, thermal limit, processor or radio parameter for Asagumo has been published in any source obtainable here, and no peer-reviewed paper on the vehicle exists: the conference abstract is the only primary source, and Spacebit’s own descriptive material is no longer online. Whether the proprioceptive, camera-free navigation approach the program described was ever implemented, in simulation or on hardware, is not documented. Because the vehicle never flew, there is no flight data of any kind to compare against the design intent, and the broader question the program posed, whether a legged CubeSat-scale rover can walk usefully inside a lunar lava tube, remains open in the published record.

References