Docking one spacecraft with another is a delicate matter. No matter which spacecraft is involved, docking requires a series of highly precise maneuvers. A slip could result in a major accident and catastrophic mission failure.
Predictably, docking without a human in the pilot’s seat is incredibly difficult. Automatic docking It is possible in Earth orbit, but if engineers really want to build autonomous spacecraft capable of docking procedures, they’ll need to give them the ability to plan their own docking maneuvers. Most spacecraft today do not have this capability. But some researchers have come up with a plan: delegate the task to an artificial intelligence system like the one behind ChatGPT.
Related: Machine learning can help track alien technology. Here’s how
Researchers have been experimenting with docking computers since the 1960s, during the early days of the space age. However, the problem has been that to correctly calculate the speed necessary for accurate docking, you need a lot of computing power. A spacecraft attempting to distance itself from Earth, or a spacecraft out of reliable contact with ground control, needs to crunch the numbers with its onboard computer — usually a supercomputer. It doesn’t happen.
“For autonomy to work billions of miles away in space without fail, we have to do it in a way that the onboard computers can handle,” he says. Simone D’AmicoProfessor of Aeronautics and one of the researchers at Stanford University Statement.
D’Amico and colleagues devised an alternative, artificial intelligence-based approach. Their method relies on transformer architecture. It’s the same type of machine learning that powers ChatGPT and many of its fellow AI chatbots. The researchers call it the “autonomous rendezvous transformer,” or ART. Instead of dealing with words like ChatGPT does, however, ART slows the spacecraft down.
The goal is for the spacecraft to run ART with its onboard hardware. It’s still early days for ART, but researchers have demonstrated its value by testing it in computer simulations. Now, they want to test it in a simulated space environment. If they can show it works there, they can get ART into orbit.
The researchers presented. Their work at the IEEE Aerospace Conference in March 2023, after a preprint was published on arxiv in October 2023.