A Robot Built Its First Truss in Orbit
Image source: Caltech/Momentus.
Gold kapton, black composite tubing, and a crowded little mechanism: from a distance the ASTRA payload looks less like a satellite than a workbench that got photographed by accident. That is roughly what it was. On the morning of March 30, 2026, the hardware rode a SpaceX rideshare out of Vandenberg Space Force Base, and for six months it worked in low Earth orbit on a task no machine had performed there before: joining a structure together out of loose parts, with no astronaut and no spacewalk anywhere near it. The machine was Caltech's Autonomous In-Space Tensioned Reflector Assembly, and the structure it was building was the perimeter truss of a 1.4-meter reflector antenna (Caltech, November 5, 2026).
Every large structure ever flown has been shaped by the inside of a rocket fairing. A mirror too big to launch has to be a mirror that folds. Astro deployable structures, the Shuttle-era solar wings, the space station's photovoltaic arrays: each trades stiffness and precision for a package small enough to pass through a payload shroud barely five meters wide. The other answer is to fly the pieces up separately and let people put them together, which is how the Shuttle assembled the space station and how Hubble was upgraded. Both routes have a ceiling, and the ceiling is the rocket. DARPA set out to attack that gap in 2022 with a program whose name is a mouthful: Novel Orbital and Moon Manufacturing, Materials, and Mass-efficient Design, or NOM4D. Its stated ambition is to ferry raw materials from Earth for in-orbit manufacturing, so that structures are designed for the environment they will live in rather than for ground test and launch survival (DARPA program page).
Program manager Andrew Detor put the scale question bluntly in the agency's November 5 announcement: "By 'very large,' I mean hundreds of meters or kilometers in scale. Nothing approaching that scale can be achieved today" (DARPA, November 5, 2026). Caltech's own reference architecture for a robotically assembled observatory is a 100-meter segmented primary, carried on more than 300 deployable truss modules that robots would build in Earth orbit before the system flew out to the Sun-Earth L2 point (Lee et al., Journal of Astronomical Telescopes, Instruments, and Systems, 2016).
The demonstration hardware came from Sergio Pellegrino's group at Caltech, which manages NASA's Jet Propulsion Laboratory, and rode to orbit hosted by Momentus aboard the company's Vigoride 7 vehicle, which carried ten payloads of its own on SpaceX's Transporter-16, a rideshare that delivered 119 payloads to sun-synchronous orbit (Caltech, November 5, 2026; NASASpaceflight, March 30, 2026). The payload itself was roughly one cubic meter. Inside it were two truss bays already built and folded for launch, an arm-like manipulator, and motorized dispensers holding the struts and joints needed for everything that followed (Caltech, November 5, 2026).
Operations began with releasing launch restraints, then pushing the two prebuilt bays out of the builder, then asking the machine to assemble a third bay from parts it had never touched. Commands came from the ground, sent by Charles Sommer, Alan Truong, and graduate student Tara Venkatadri, and the only way to know whether any of it had worked was to download photographs from the payload's eleven cameras and compare them against what should have happened (Caltech, November 5, 2026). Anomalies were plentiful. The workaround was to re-create the failure on a flight-spare copy of ASTRA sitting in Pellegrino's lab, solve it there, and send the revised recipe up. There were only about seven or eight contact windows a day, each roughly fifteen minutes long, so the machine spent twenty-two hours of every twenty-four on its own. Six months in, the antenna was not finished. The truss was designed for eleven bays, and the demonstration had added one.
Two design decisions explain why it worked. The first is the manipulator. DARPA describes it as a gantry robot able to move along three linear axes, and that constraint does more work than it first appears (DARPA, November 5, 2026). A free-flying arm in orbit has to hold its own pose against a moving vehicle, and place a part accurately while reacting against the very structure it is building. Bolted to a rail, the arm inherits its position from the machine carrying it, and a hard control problem collapses into repeatable Cartesian motion. The gantry is not clever. It is a decision to spend the complexity on the joints instead of on the arm.
The second decision is the surface itself. The reflective layer ASTRA deploys is a perforated polyimide film, Kapton, co-invented with a group at the University of Pennsylvania led by Igor Bargatin, and it functions as two cable nets prestressed against one another by springs (Caltech, November 5, 2026). That geometry matters because a tensioned net holds its shape through tension rather than through stiffness. For a cable spanning a distance L under a distributed load q, the out-of-plane sag runs roughly as δ = qL²/(8T), with T the tension per unit length. Halve the tension and the sag doubles; a structure needing less prestress can therefore be built with less material holding it up. Knitted metallic mesh, the standard in furlable reflectors, asks a lot of prestress. Perforated film asks less, so the truss behind it can be lighter. In a program whose premise is that a kilometer of structure must stay affordable to launch, shaving prestress is not a detail. It is the budget.
Scale the demonstration up and the numbers get interesting. Eleven bays span the perimeter of a 1.4-meter aperture, a circumference of roughly 4.4 meters, or about two and a half bays per meter of edge. Carry that density to a 100-meter aperture and you need some 800 meters of truss, a few hundred bays, which is the scale the RAMST reference design assumes (Lee et al., 2016). Put the aperture against the diffraction limit, θ ≈ 1.22λ/D, and a 100-meter mirror working at 500 nanometers resolves roughly 1.3 milliarcseconds, against about 50 milliarcseconds for Hubble. That is the prize, and it is why the concept splits the observatory into parts: the primary mirror, an optics and instrumentation unit a few hundred meters away, a control unit beyond that, and a spider-like six-armed robot, a hexbot, that assembles the truss and then walks across it to install the mirror segments.
What ASTRA established is narrower than the headline and more useful. One bay, built autonomously, in six months, with a human verification loop running about two hours a day, proves the process rather than the product. The complementary half of NOM4D runs in parallel: the other Phase 3 team, at the University of Illinois Urbana-Champaign, uses a station module to harden carbon tubes in orbit, making the material up there instead of merely putting it together (Caltech, November 5, 2026). ASTRA's mission winds down in December, with the team already pointing at the next round of simplification: fewer failure modes, less ground support, shorter sequences.
The open question is whether the limit on building in orbit is dexterity or bandwidth. ASTRA suggests bandwidth. The robot could pick and place, but every fifteen-minute window had to carry diagnosis, redesign, and a retest, and better hardware does not erase that arithmetic. The fairing has always been the ceiling on what we can put in space. The next twenty years will test whether a gantry, a box of tubes, and a few thousand contact windows a year can lift it.
Featured blog posts
A Transformer model called EarlyDetect, trained on SDO/HMI helioseismic and magnetic observations by an NJIT-led team including Princeton and NASA Ames, forecasts the emergence of solar active regions on the Sun an average of 9.24 hours before they become visible, outperforming LSTM baselines and revealing that temporal smoothing filters hid the faintest precursor signals (Tirona et al., JGR Machine Learning and Computation, 14 August 2026) Posted on May 24, 2026
Posted on May 4, 2026
Today we are joined byYasunori Yamazaki, Chief Business Officer at Axelspace. Axelspace are pioneers of microsatellite technology advancing the frontiers of space business, reimagining traditional ways of using space, and creating a society where everyone on our planet can make space part of their life. Latest blog posts
For six months, a Caltech robot on a Momentus vehicle worked through a box of tubes and joints in low Earth orbit, and managed to add a single bay to a truss. The result says less about robot dexterity than about the ground loop: with only seven or eight fifteen-minute contact windows a day, verification bandwidth, not manipulation, is what limits building in orbit.
Mojgan Aghakhanloo and colleagues at the University of Virginia used a five-epoch Chandra Target of Opportunity campaign in March 2026 to make the first X-ray detection of the supernova impostor AT 2016blu, a luminous blue variable of at least 33 solar masses in the spiral galaxy NGC 4559 about 30 million light-years away, and found it 100 times brighter in X-rays than the 2001 and 2002 archival non-detection, with a hard spectrum and a luminosity of log L_X = 38.64 erg/s that only accretion onto a hidden neutron star or stellar-mass black hole can explain, the first luminous blue variable found with a compact companion
"Menlo Research's Asimov 1 is a 1.2 m, 35 kg, 25-joint humanoid whose mechanical CAD, electrical design, MuJoCo and URDF simulation models, firmware, reinforcement-learning training code and Python SDK are all published under CERN-OHL-S-2.0, GPL-2.0 and BSD-3-Clause, and by early November 2026 outside developers were submitting work against it, including a runtime safety layer experiment and a pull request adding the robot to the LeRobot framework"'
Liu, Saper and colleagues report in Science that Teghaza 001, a 798.5 gram Martian meteorite recovered in Mali in 2022, is a slowly cooled igneous rock formed at least 4.1 billion years ago that samples an unsampled part of Mars' early crust and preserves a chemical record of water rich fluids and of hydrogen lost to space
A team led by Anna Karpova and James Taylor used the Green Bank Telescope to find PSR J2053+6905, a 1.68 millisecond binary pulsar and the third fastest redback known, whose optical light curve brightened on one side between 2023 and 2025 and whose mass of 2.12 solar masses constrains the equation of state of ultradense matter.
NASA's announcement gives a takeoff weight of 10,000 pounds, a Mach 0.75 cruise and room for 2,000 pounds of swappable experiment, but leaves out the engine count, the wing geometry and the contractor. What the airframe is actually for is finding out how much control authority a crewed airplane can be given before the ordinary surfaces stop working.
United Launch Alliance's Vulcan Centaur suffered an anomaly 10 minutes into its launch on 15 November 2026 from Space Launch Complex 41, destroying 40 Amazon Leo satellites on the inaugural Leo Vulcan 1 mission and marking ULA's first total mission failure since the Boeing and Lockheed Martin partnership was formed in 2006, with the failure occurring during the hydrogen-burning Centaur upper stage's planned 12-minute 15-second burn after the solid boosters worked normally and the Blue Origin BE-4 first stage appeared to burn longer than its pre-launch timeline expected
"Protopapa et al. 2026 in The Astrophysical Journal Letters report that JWST/NIRSpec spectra of interstellar comet 3I/ATLAS taken in August 2025, December 2025 and April 2026 show the coma's water ice locked into sub-micron aggregates loaded with rocky material, which suppresses the 1.5 and 2.0 micron water-ice overtones while keeping the 3 micron band deep, placing 3I/ATLAS at the dusty end of a continuum of ice grain states shared with trans-Neptunian objects"'
SpaceX caught the Starship upper stage Ship 42 between the arms of Orbital Launch Tower 2 at Starbase at 21:20:03 UTC on 14 November 2026, the first catch of an orbital rocket upper stage returning from space, after 22 hours 32 minutes and 14 orbits in which the ship deployed 26 Starlink V3 satellites and was photographed in orbit by three of them, while Super Heavy Booster 22 lost its own catch when a low fuel mass alarm ended the boostback burn early
Lithium fluoride crystals flown inside two human-shaped phantoms on the 2022 Artemis I mission show that radiation dose near the Moon exceeds low Earth orbit by less than expected, because the vanishing Van Allen proton contribution is offset by galactic cosmic rays that pass through a body. The dose also comes out nearly uniform through a torso, which changes how lunar shelters have to be built.
SpaceX's fifteenth Starship flight lifted off from Starbase at 22:47 UTC on 13 November 2026 as the first attempt to catch both stages of an orbital rocket with the launch tower's mechanical arms, but Super Heavy Booster 22 was directed to an offshore splashdown in the Gulf of Mexico about six and a half minutes after liftoff instead of being caught, while Ship 42 reached a 275 km orbit, deployed all 26 Starlink V3 satellites, and began a roughly 21 hour coast before a first ever attempt at catching a Starship upper stage at the tower, scheduled for about 21:20 UTC on 14 November
Subscribe to blog posts using RSS










