Research results show that the synthetic intelligence (AI) architecture, which is the idea of 'ChatGPT', will be used for docking tasks that match the orbits and adjust the speed to attach the entrances and exits in order that spacecraft flying at greater than 7 kilometers per second don’t collide. got here out.
Space.com reported on the fifteenth (local time) that Stanford University's aerospace research team has unveiled 'ART (Autonomous Rendezvous Transformer)', an optimal orbit creation model for docking two or more spacecraft using AI.
In response to this, as an alternative of learning and generating language like ChatGPT, ART analyzes spacecraft trajectories. It’s because the Transformer architecture, which relies on ChatGPT, may be very efficient in parsing various kinds of data similar to images, audio, and trajectories in addition to words.
The core of ART is to quickly generate high-quality trajectory candidates by integrating AI into the prevailing spacecraft trajectory calculation pipeline.
“One in all the largest challenges on this field is the ‘ground within the loop,’ which requires supercomputers to relay information to the bottom where they calculate orbits, after which upload commands back to the satellite,” said Simone D’Amico, associate professor of aeronautics and astronautics at Stanford University. )’ problem,” he said, emphasizing, “If you should operate autonomously billions of miles away in space, you may have to do it in a way that the onboard computer can handle.”
Spacecraft attempting to dock too removed from Earth or unable to contact ground control cannot use a supercomputer, and must as an alternative run ART with their very own onboard hardware.
The researchers claimed that although ART isn’t the primary AI model to be applied to a spaceflight problem, it outperformed other AI-based models in computer simulation tests conducted in a terrestrial laboratory environment.
The following step is to check it in a simulated space environment. If this test is deemed acceptable, ART can go into space orbit.
“Our next step is to enhance the functionality of ART by infusing it with additional AI elements,” Associate Professor D’Amico said. “In fact, there remains to be a protracted strategy to go before we will test ART in space.”
Reporter Park Chan cpark@aitimes.com
