Every large object in orbit today got there as a payload folded to fit inside a rocket fairing. A small Caltech experiment that spent six months in space just tested a different approach. Instead of unfolding a finished structure, it built one there, piece by piece, with a robot arm and no human hands guiding it.
In statements dated Nov. 5, Caltech and DARPA said the Autonomous In-Space Tensioned Reflector Assembly, or ASTRA, had autonomously assembled the first section of a reflector antenna structure in orbit. "This is the first time that a structure has been autonomously constructed in space," said Caltech's Sergio Pellegrino.
That claim has a firm limit. The structure was meant to support a 1.4-meter-diameter reflector antenna, and ASTRA ran out of time before finishing it. Only the initial section got built. According to Caltech, that was still enough to show that the concept is feasible and can scale.
What actually flew
ASTRA is a truss builder. It assembles a structure's frame, which is not the same job as stacking blocks. Its target was a perimeter truss: a ring made of identical repeating units called bays, which would hold a reflective surface under tension. The parts were composite fiber tubes for struts, aluminum joints to connect them, and Kapton for the reflective surface.
To build each bay, an autonomous manipulator picks struts and joints out of dispensers and puts them together. Space.com reports that the payload flew with two prebuilt truss sections, the robotic manipulator and a supply of parts. The robot's job was to add new structure onto that starting hardware.
The operating conditions were tight. According to Space.com, communication windows totaled about 15 minutes a day, and engineers used them to send commands and check on the robot's progress as it picked up parts, readied them in the right spot and built on an assembly plate. With so little contact, operators can't remotely steer every strut into place, which is much of what "autonomous" means here. The team overcame many "little surprises" along the way but could not finish the entire truss in time.
ASTRA did not have its own spacecraft. It rode as a hosted payload on Momentus's Vigoride 7 space tug, which launched March 30 on SpaceX's Transporter-16 rideshare mission. Caltech says the payload was completed, tested and delivered to Momentus in under 12 months.
The program behind it
ASTRA is part of DARPA's NOM4D program, short for Novel Orbital and Moon Manufacturing, Materials, and Mass-efficient Design. Caltech says DARPA started NOM4D in 2022. ASTRA is the third phase of the project, and Caltech says it is one of only two projects to get that far. The first two phases developed the concept and design methodology, then showed in the laboratory that it worked.
DARPA's goals go far beyond a 1.4-meter dish. "The goal is very large structures, hundreds of meters or kilometers in scale," NOM4D program manager Andrew Detor said, according to Space.com. "Nothing approaching that scale can be achieved today."
The other phase 3 project comes from the University of Illinois Urbana-Champaign and takes a very different route. Instead of snapping together prefabricated parts, the Illinois team works on making structural material in orbit. Universe Today describes a self-propagating reaction in which a heated liquid monomer forms a hardened polymer, creating long structural tubes. NOM4D has not yet performed this process in space. Space.com reports the project will go to the International Space Station to study how carbon tubes harden in orbit.
Together, the two projects cover two parts of the problem. One is how to assemble parts autonomously. The other is how to make the parts themselves in orbit from raw material.
Why It Matters
Two familiar constraints limit how big a space structure can be. The first is the fairing. Anything launched has to fit inside the rocket's nose cone, so large antennas and reflectors either stay small or get elaborate folding mechanisms, which add mass, complexity and ways to fail. Universe Today notes that telescopes such as JWST use ingenious folding designs, but that this is not practical in every case. The second constraint is the rocket equation, which Universe Today says refuses to bend to our desires: a heavy, rigid structure built to survive launch is costly to lift.
Assembling in orbit gets around both. Smaller components can be sent up inside fairings and assembled into a larger structure once there, as Universe Today puts it. A robot that adds identical bays over and over can, in principle, keep going as long as the parts last. That is the scalability Caltech points to. In space, a truss twice the size needs twice the parts, not a fairing twice as large.
The program sits in this publication's military category for a reason. DARPA is funding it, and Detor's talk of kilometer-scale structures points toward very large antennas and reflectors. Neither Caltech nor DARPA has tied ASTRA to a specific mission, and the demonstration itself was modest: part of a ring meant to hold a 1.4-meter reflector. Caltech's photo captions say the truss builder launched with two prebuilt bays and built a third in orbit.
Still, the demonstration matters. In the first two phases, Caltech showed the concept only under laboratory conditions on Earth. ASTRA showed that a robot can pull parts from dispensers and join them into structure in orbit, working mostly unattended with about 15 minutes of contact a day. Whether the method can grow from a partial 1.4-meter truss to the hundreds of meters DARPA wants is still unproven. What has changed is that it now has a successful test in orbit to build on. Detor himself called the demonstration "not an endpoint; it's a beginning."
