NASA’s iSAT study. A couple years ago, NASA’s Astrophysics Division spent two years assessing the feasibility of assembling a large-aperture observatory in space. The In-Space Astronomical Telescope (iSAT) Assembly Design Study (pdf) concluded that In-Space Assembly (ISA) is the only option for building observatories with aperture diameters >15 m and would even be likely to be beneficial for smaller ones like the JWST (6.5 m aperture). Efforts like Northrop Grumman’s successful Mission Extension Vehicles, the upcoming DARPA RSGS mission, and NASA OSAM-1, along with the usage of Canadarm2 to install instruments with standardized interfaces on the outside of the ISS, all demonstrate the increasing maturity of robotic servicing and assembly. The iSAT study describes a telescope composed of modules with standardized interfaces, launched with a spacecraft bus that has attached Canadarm2-like robotic arms that can assemble and deploy modules delivered by space tug from multiple launches. The benefits over launching monolithic spacecraft with hundreds of single points of failure (cough JWST cough) are clear: the mission won’t be limited by a single launch vehicle’s lift ability or fairing size; the same inchworming robotic arm that does initial ISA can later perform repairs and upgrades, either with freshly delivered replacement modules or by debugging malfunctioning parts (see Mars Insight); the final deployed structure doesn’t need to be designed to handle the harsh launch environment; and, design and development will be faster without needing to design and test super reliable deployment mechanisms—if a part fails during orbital checkout, launch a replacement. The primary challenge is designing hardware that today’s limited-dexterity robotics can manipulate, and figuring out supervised autonomy with fallback telerobotics for bringing humans into the loop when needed. There are definitely challenges, but this feels like a feasible approach. If one could do it near a crewed station for infrequent debugging EVAs, even better.