\n<\/p>\n
Fly to an asteroid, land on it, make no mistakes: If only it were that easy. <\/p>\n
When NASA sends spacecraft to explore the solar system, they build in redundancy and automation. But much of the work is done by teams of flight controllers communicating with the vehicle from Earth. The Osiris-Rex mission, which rendezvoused with an asteroid in 2018, had 100 operators on each eight-hour shift.<\/p>\n
But resources like that aren\u2019t available to startups, so AstroForge, which is developing technology to mine asteroids, is turning to AI. The company has developed an autonomous control stack for its spacecraft called \u201cSolo,\u201d a transformer-based model made in-house.<\/p>\n
AstroForge plans to fly its first autonomous spacecraft in 2027 on the first rocket launched by Stock Space. That mission will be backed by NASA and is expected to gather scientific data about the Sun.<\/p>\n
That would be an accomplishment: Most spacecraft autonomy depends on traditional control algorithms due to concerns about the unreliability of neural networks. The first use of a neural network to control a satellite\u2019s positioning in orbit took place just last year<\/a>.<\/p>\n AstroForge, founded in 2022 and backed with $56 million in venture funding, has launched two prototype spacecraft, both of which suffered anomalies that prevented them from achieving most of their mission objectives. In 2025, its Odin spacecraft was launched into deep space, but the company had difficulty communicating with it. There are a limited number of antennas on Earth big enough to transmit to spacecraft hundreds of thousands of miles away, and the windows of time in which to do it are small.<\/p>\n Ultimately, AstroForge couldn\u2019t gain control of Odin, but the experience spurred it to consider alternatives: Could it put sufficient intelligence onboard the spacecraft for it to solve its own problems?<\/p>\n \u201cWould that have been recoverable with all the data on the spacecraft? I don\u2019t know, but I can tell you nothing onboard tried it, and I would love something onboard to try if the spacecraft is unrecoverable at launch,\u201d Matthew Gialich, AstroForge\u2019s co-founder and CEO, said. <\/p>\n \u201cThe trade for me is: Do I go build my own ground network, which is going to cost [around] $200 million to put up five dishes around the world and then do operations on it, or do I try to remove it with a model?\u201d<\/p>\n Armand Awad, AstroForge\u2019s head of flight software, said the company decided to leverage the advances in transformer models driven by the frontier labs. It created a stack that includes traditional control algorithms, models trained on test data for specific subsystems like power generation or navigation, and an overall intelligence layer trained on about 2,500 sensors in the spacecraft.<\/p>\n \u201cI\u2019m not saying I\u2019m going to make general spacecraft autonomy or general autonomy for the world,\u201d Gialich said. \u201cI\u2019m making a constrained autonomy at a very low sensor input, following the basic training of a transformer model.\u201d<\/p>\n In theory, the agent in control of the spacecraft will perform tasks like anomaly resolution. Awad imagines it realizing that it has lost track of its position in space, correlating a power anomaly to issues in its star tracker, and fixing the whole thing \u2014 \u201cprobably turning it on and off in this case.\u201d<\/p>\n The company\u2019s third vehicle, DeepSpace-2, is currently set to launch alongside Intuitive Machines\u2019 third moon mission, expected to head for space by the end of 2026. Solo will be onboard that vehicle, flying in \u201cshadow mode,\u201d so AstroForge\u2019s engineers can put it through its paces before the Autonomy-1 mission.<\/p>\n Will this truly be an AI agent operating an independent spacecraft?<\/p>\n \u201cI don\u2019t plan on flying radios that can receive from Earth on Autonomy-1,\u201d Gialich said. \u201cWe have to go all in right now. The team\u2019s probably going to talk me into it by the time we fly it. But right now, I\u2019m telling them no radios.\u201d<\/p>\n<\/div>\n When you purchase through links in our articles, we may earn a small commission<\/a>. This doesn\u2019t affect our editorial independence.<\/em><\/p>\n