A robot built a truss in orbit, after a 15‑month sprint
Picture a metal lattice slowly taking shape while the Earth rolls past beneath it: no astronaut with a wrench, just a boxy robot sliding along rails, clicking struts into place. That’s the scene behind a quiet first—an automated assembly carried out in orbit, not on a lab bench. NOM4D says it has reached this milestone, and even its own team did not expect it so soon.
Why now? In February 2025 the program tore up its timeline and aimed straight for space, and by March 2026 SpaceX had launched Caltech’s NOM4D payload. The program’s manager, Andrew Detor, puts the sprint in plain terms: they went from the idea of an on‑orbit demo to a working prototype in 15 months, and testing in the real environment is how you mature a space technology fast.
From factory floors to freefall: what a gantry robot does up there
The demonstrator used a gantry robot—think of the three straight-line rails over a CNC mill—tasked with building a truss. A truss is the engineer’s go‑to skeleton: light pieces joined into a stiff framework. On Earth, gantries thrive because they’re simple, predictable, and precise; in orbit, that simplicity is an asset when every moving part is a liability.
Analogy: imagine a 3D printer that lays down bars instead of plastic. Where it breaks down is that the space version can’t push against a table—every action nudges the robot in microgravity. The kit has to manage its own reaction forces and keep itself metered to a reference so the lattice grows straight rather than wandering off-angle.
Why start with a truss?
NOM4D’s objective is to make construction in orbit efficient and scalable for truly large structures—Detor defines “very large” as hundreds of meters to kilometers. Nothing close to that size can be launched intact today, so an architecture that repeats a simple operation many times is the fastest path to scale. A truss is modular by nature and forgiving of small local errors while still demanding overall alignment—exactly the right training ground for a first automated build.
It’s also a test of process, not just parts. If you can reliably join members at nodes, verify the geometry, and move the work out of the assembly box, you have the foundations of a construction line in orbit. That’s more valuable than any single demo artifact because it’s the part that scales.
Autonomy without a safety net
“Operating in space is much harder than operating on the ground” was Detor’s understated warning: there’s virtually no margin to rescue a jammed end-effector or a misaligned joint. Automated space assembly has to anticipate faults, detect them quickly, and either correct or gracefully abort before a bad piece becomes embedded in the structure.
That means conservative motion plans, robust sensing of relative positions, and simple mechanisms that don’t depend on delicate timing. It also argues for building inside a controlled “box” first—contain the work, constrain the degrees of freedom, and then deploy finished sections once they’re verified. Fewer clever moves, more solid engineering, as the team emphasizes.
Making the parts on site: resins and curing in orbit
Assembly is step one; making bigger, stronger elements in space is step two. Next up, NOM4D says a team from the University of Illinois Urbana‑Champaign will fly a test to the International Space Station to demonstrate composite material processing. The plan is to use a monomer resin and, with an energy‑efficient heating method, convert it into a strong polymer—akin to how aerospace composites are made, but done off Earth.
If you can cure structural members on orbit, you’re no longer bound to launching every beam and panel. You can ship feedstock and tools, then form parts at full scale where they’ll be used. That’s a direct lever on the size ceiling Detor wants to lift.
The hard limits: alignment, power, and heat
Three constraints dominate early orbital construction lines. First, metrology: you need a stable reference to keep assemblies square over many joints. Drift accumulates. Second, power: even “energy‑efficient” heating and continuous robotics draw steady watts, which must be scheduled and stored. Third, thermal swings: components cycle between hot and cold every orbit, so joints and cured parts must tolerate expansion and contraction without creeping out of spec.
These are solvable, but they dictate architecture. Build in sections you can verify, use joints that tolerate tiny misfits without play, and design processes that don’t depend on narrow thermal windows. The early choice of a simple gantry and truss points in exactly that direction.
The deeper cut
Momentum management for orbital assembly
On a gantry in microgravity, every placement force injects linear and angular momentum. Without ground reaction, the robot must close its momentum budget internally: counter‑moving axes, reaction wheels, or controlled contact with a base. If an axis accelerates a 1 kg toolhead at 0.1 m/s² for 0.5 s, it imparts 0.05 N·s; unless another axis or wheel provides −0.05 N·s, the carriage will drift, biasing the next placement. Over many operations, uncompensated impulses accumulate into millimeters of error, which at truss scale becomes degrees of misalignment. That’s why build‑inside‑a‑fixture is attractive: contact constraints convert free‑drift into constrained motion with known reaction paths, easing controller design. For in‑situ curing, polymerization adds its own mechanics: exotherm management and shrinkage strain. A resin that shrinks, say, a fraction of a percent upon cure can preload a joint or pull a long member out of straightness unless the cure is sequenced symmetrically and heat is drawn down gradients carefully. The engineering trade is between higher‑modulus networks (stronger, stiffer) that often have higher shrinkage and lower‑shrink systems that may underperform structurally; “energy‑efficient” heating suggests localized cure fronts, which helps with peak power but complicates stress management.
What happens when it fails?
In orbit, failure containment is design work, not emergency work. A stuck connector or mis‑seated member should trigger a halt before the next step, with the subsystem able to back out to a known state. Redundancy favors duplicated simple end‑effectors over one complex hand; graceful degradation beats heroics.
Because rework is expensive, the process is the product: alignment checks baked into every placement, and section‑by‑section verification before deployment. That’s how you keep a local fault from becoming a structural defect meters later.
What we don’t know yet
No one has publicly put numbers on the largest structure NOM4D could assemble today, the cost per kilogram for on‑orbit fabrication, or the endurance of resin‑cured parts after long exposure. The recent on‑orbit demonstration shows feasibility, not final performance. As Detor puts it, the Caltech demo is a beginning, and the real question now is what to make once a process that can scale to large structures has been shown.
The fork in the road: build what?
NOM4D’s own framing is clear: the goal is efficient, scalable construction of very large structures, and nothing at the hundreds‑of‑meters‑to‑kilometers scale is achievable by launch alone today. With an assembly line that worked once in orbit and a materials experiment queued for the ISS, the next choice is architectural—what structure justifies the complexity and benefits most from size?
Detor’s challenge is implicit: now that a path to scale exists, decide what capabilities to unlock with it. That’s a technology question first, a mission question next.
Sources: Automated assembly in space is the future and NOM4D is showing the way (phys.org)
Images: Cover: Everyman Science (AI illustration)
How this article was made: Everyman Science uses AI tools to structure, format and optimise its articles, and occasionally to produce illustrations where no free photograph exists. The reporting these articles are based on is human-produced and cited above. Spotted an error? Write to [email protected] and we will correct it. — The editors How we work.
Science desk team of Everyman Science, curating and reporting on the day’s most significant developments in research, space exploration, and technology.
