Autonomous Spaceflight: AstroForge's AI-Driven Revolution

From Earth Control to Autonomous Navigation

When NASA sends a spacecraft into deep space, it's a meticulously orchestrated dance of engineering and communication. Thousands of flight controllers monitor missions like OSIRIS-REx, which spent years orbiting an asteroid and collecting samples. That kind of precision requires extensive ground support — not to mention the logistical challenges of maintaining real-time communication across vast distances.

But what happens when you're a startup with limited resources? Enter AstroForge, a company developing technologies for asteroid mining and space exploration that's betting on artificial intelligence to solve these challenges.

"Would that have been recoverable with all the data on the spacecraft? I don't 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," said Matthew Gialich, CEO of AstroForge.

Gialich's point underscores a fundamental shift in how we think about spaceflight. Rather than relying on continuous Earth-based control, AstroForge aims to give its probes an onboard intelligence capable of making decisions — even in the event of a communications failure.

Building AI into Spacecraft

The heart of AstroForge's innovation is a new software stack called Solo, designed specifically for spacecraft autonomy. Built using transformer-based neural networks — models that have revolutionized natural language processing and are now being adapted for complex control tasks — Solo allows the craft to interpret sensor data, diagnose anomalies, and respond autonomously.

It's not just about replacing human operators with machines. Solo is engineered with constraints in mind, designed to work within a limited set of inputs. As Armand Awad, AstroForge's head of flight software, puts it, "I'm not saying I'm going to make general spacecraft autonomy or general autonomy for the world. I'm making a constrained autonomy at a very low sensor input, following the basic training of a transformer model."

This approach addresses one of the core concerns in space exploration: reliability. Neural networks are still relatively untested in the harsh environment of outer space. The first use of neural networks for satellite control only occurred last year — and even then, it was limited to positioning tasks. AstroForge's effort marks a step forward by integrating AI across multiple subsystems, not just navigation.

Learning from Failure

AstroForge's journey hasn't been without setbacks. Its first two prototypes experienced anomalies that prevented them from fulfilling their mission objectives. Most notably, the Odin spacecraft launched in 2025 encountered communication problems, leaving it effectively out of reach. It was a wake-up call.

"The 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?" Gialich asked. That question became the catalyst for a new approach — one that shifts control from Earth to the craft itself.

The company has since pivoted to developing an AI-driven solution that could handle issues in real time, reducing reliance on expensive and time-sensitive communication infrastructure.

Autonomy-1: The Next Milestone

That approach is now coming to a test flight in 2027 with the launch of AstroForge's Autonomy-1 mission. Scheduled for liftoff on a rocket from Stoke Space, this mission will be backed by NASA and is expected to gather scientific data about the sun.

But what makes Autonomy-1 truly groundbreaking isn't just its mission — it's how it will operate. Unlike past missions that depend heavily on ground-based command, Autonomy-1 will fly with no communication radios. That means it's designed to function completely independently, relying entirely on onboard systems to navigate, make decisions, and even recover from potential failures.

"I don't plan on flying radios that can receive from Earth on Autonomy-1," Gialich said. "We have to go all in right now. The team's probably going to talk me into it by the time we fly it. But right now, I'm telling them no radios."

This is a bold move, one that pushes the boundaries of what's possible with current AI and spacecraft engineering. If successful, it could mark a new era in space exploration — one where probes don't just send data back to Earth but act like autonomous agents exploring unknown territory.

Implications for the Future of Space Exploration

The implications of AstroForge's work extend beyond their own mission. As AI models become more sophisticated and compact, we're seeing a broader trend in space technology — the move toward smarter, more autonomous systems. This is especially critical as private space ventures proliferate and the cost of launches continues to decline.

This shift is also tied to economic considerations. The expense of maintaining global communication networks for deep space missions can be prohibitive for startups and smaller organizations. By embedding AI directly into spacecraft, AstroForge offers a path forward that could democratize access to space exploration — even if it's still a long road ahead.

Moreover, this work contributes to the larger conversation about AI's role in complex environments. While many AI applications today are focused on data processing or pattern recognition, AstroForge is showing how these models can be used for real-time decision-making under high-stakes conditions — a critical leap in the evolution of artificial intelligence.

Looking Ahead

AstroForge's journey to autonomous spaceflight is just beginning. Their next vehicle, DeepSpace-2, is scheduled to launch alongside Intuitive Machines' third moon mission by late 2026. While it will operate in “shadow mode,” this mission will serve as a crucial testbed for Solo before the real challenge of Autonomy-1.

For now, AstroForge's work stands as a testament to how innovation in space exploration can be driven not just by hardware but by software — specifically, by the integration of artificial intelligence into the very core of spacecraft systems. Whether they succeed or not, their bold approach will shape the conversation around what it means to explore the cosmos in an era where AI is increasingly a part of our lives.

As we continue to watch how these technologies evolve, one thing remains clear: The future of spaceflight isn't just about launching more rockets — it's about building smarter spacecraft that can think and act independently, guided by artificial intelligence rather than just the commands of engineers on Earth.