Caltech robot autonomously Assembles truss bay in orbit

A Caltech-developed robot autonomously assembled a truss bay in low Earth orbit after launching March 30, 2026, aboard Momentus Inc.’s Vigoride 7 orbital vehicle from Vandenberg Space Force Base. The vehicle carried 10 separate payloads into orbit on the SpaceX Transporter-16 rideshare, and the ASTRA payload occupied roughly one cubic meter, equal to 35 cubic feet. Caltech reported the first autonomous in-space assembly of a structure without astronauts on November 9, 2026.

ASTRA aimed to assemble a reflector antenna spanning 1.4 meters, or 4.6 feet, over six months in orbit. The project, known as the Autonomous In-Space Tensioned Reflector Assembly, relies on composite fiber tubes, aluminum connectors, and a Kapton film. Its complete design calls for 11 connected bays that extend outward from the assembly system.

Before launch, the payload carried two truss bays assembled on the ground. They included diagonal tubes and connecting joints, and the machine also held motorized dispensers with parts for the remaining sections. After extending the prebuilt sections, the robot built one more bay. The mission will end in December without finishing all 11.

The robot’s automated arm takes structural tubes and joints from motorized dispensers. It joins them into truss bays on an assembly plate, then pushes finished sections outward into space. After the robot finishes a bay and its reflective parts are ready, the system pushes the completed section outward. It then begins assembling the next bay and repeats the sequence.

Designers built the truss from composite fiber tubes, aluminum connectors, and a lightweight Kapton reflective surface. Once the robot assembles the bay framework, it adds a thin reflective film with many openings. The film forms two reflective cable nets held under tension by springs to keep the surface in shape.

The perforated Kapton film needs less tension than the knitted metallic mesh normally used in foldable reflectors. Engineers can use less supporting structure and make the assembly lighter. Lower weight matters because each added kilogram raises demands on a launch vehicle.

Ground controllers had only about seven or eight daily contact windows, each lasting roughly 15 minutes. Charles Sommer, Alan Truong, and Tara Venkatadri, three team members, sent construction commands through the Vigoride vehicle and checked photos from orbit.

Commands went from Earth to the onboard computer through the Vigoride spacecraft. The team checked results using photographs downloaded from orbit. Engineers used a second ASTRA system kept in Sergio Pellegrino’s Caltech laboratory as a flight spare. They recreated in-orbit problems on the ground and tested possible fixes before sending revised instructions.

Caltech researchers built the robot using DARPA funding through the NOM4D program, which DARPA started in 2022. During the first two program stages, researchers developed the construction concept, worked out the engineering design, and verified operation in Earth-based laboratory experiments. The team had less than 12 months to build, test, and deliver the hardware to Momentus Inc., which hosted the payload. The quick delivery schedule built on the Earth-based concept and design work from the program’s first two stages.

Sergio Pellegrino, a Caltech professor and senior research scientist at JPL, led ASTRA. He also co-directs Caltech’s Space Solar Power Project. Caltech manages JPL for NASA.

Caltech’s Space Solar Power Project seeks to gather solar power in space and transmit it wirelessly to Earth’s surface. Caltech and University of Pennsylvania researchers led by Professor Igor Bargatin co-invented the perforated Kapton sheet used as the reflector film.

Jaden Suh, now an assistant professor at the University of Texas at Austin, was a former Caltech postdoctoral scholar. He helped build the system from scratch and described sending his work to space as a lifelong dream.

Andrew Detor, a DARPA program manager, said NOM4D targets structures spanning hundreds of meters or kilometers. He said no existing approach can achieve that scale today. DARPA imagines eventual structures several football fields long or even extending for kilometers. They would far exceed what conventional spacecraft or current launch methods can achieve.

DARPA likens the assembly process to building with advanced versions of K’Nex or Tinkertoy construction toys. Automating those same steps reliably in space poses a far greater engineering challenge than the toy comparison suggests. DARPA’s long-term vision includes robots building large radio antennas, solar power stations in orbit, other orbital infrastructure, and structures to support human missions beyond Earth.

Robotic construction provides a third path past rocket fairing limits: send materials to orbit and assemble systems there rather than folding them or relying on astronauts. In the past, engineers folded structures for launch and unfolded them in orbit, or they launched components for astronauts to assemble. Both approaches cap the size of the finished structure.

Bigger antennas could deliver communications capabilities that conventional deployable spacecraft cannot practically provide. Bigger telescope reflectors could gather more light and observe more distant objects than current launch fairings permit.

Caltech’s report says assembling large structures directly in space could transform astronomy, communications, and space-based energy systems. It would enable structures larger than rocket fairings allow. Getting rid of that limit could cut reliance on pre-launch folding and astronaut spacewalks for station construction and other orbital infrastructure.

The experiment did not demonstrate construction at enormous scales. Its achievement was proving that a robot could assemble structural components in orbit, an essential step toward much larger systems.

Only two NOM4D phase-three projects exist, and ASTRA is one of them. A University of Illinois Urbana-Champaign team leads the other project. It will test carbon-tube hardening aboard the International Space Station. The program’s first two phases developed the concept and verified it in Earth laboratories before moving to the orbital demonstration.

Researchers do not expect ASTRA to finish all 11 bays before the mission ends in December. They still consider feasibility demonstrated.

The experiment revealed ways to simplify and improve the robotic hardware and assembly procedure. The team says it has learned enough lessons to show the on-orbit assembly approach is viable. Future work will focus on simplifying and refining the design.