New Spacecraft Autopilot Lets Tiny Satellites Dock Themselves

Docking two spacecraft in orbit is one of the trickiest maneuvers in spaceflight. A chasing satellite has to constantly adjust its position and speed, and small forces like solar pressure can push it off course. Most modern controllers solve an optimization problem in real time: they look at the current state, calculate the best next move, then repeat. That works, but it has two big drawbacks. The calculation time can vary and sometimes take too long, and there's no formal guarantee that the whole system will stay safe in every possible situation.

This paper offers a smarter approach. Instead of solving the math on the fly, the researchers precompute the solution and store it as a lookup table of simple rules. Previously, that idea was considered impractical because the table grows explosively as the scenario gets more complex. But for spacecraft rendezvous, the authors found special structures in the problem that keep the table compact. At a 51-step horizon, the entire control law takes only 10.5 MB and runs in under 300 nanoseconds—fast and small enough to fit on the modest processor used in a CubeSat, a satellite about the size of a shoebox.

Just as important, the controller isn't tuned by trial and error. The team used reachability analysis, a technique that computes every possible state the spacecraft could reach, to verify offline that the system remains stable even with bounded disturbances. That gives engineers a mathematical proof of safety instead of relying only on simulations. For spacecraft certification, that's a huge deal.

The practical effect is that small, cheap satellites could one day perform autonomous docking, refueling, or debris removal without needing a powerful flight computer or constant human oversight. The catch: the results assume known limits on disturbances and haven't yet been tested on a real mission. Still, this is a step toward safer and more affordable space operations.