AstroForge is advancing space mission autonomy with its onboard transformer AI, Solo, set to command its next spacecraft without ground radio support, marking a significant shift in small satellite operations.
- AstroForge uses AI autonomy to manage spacecraft operations.
- Solo AI integrates sensor data to handle anomalies and navigation.
- The next mission will test AI control without Earth radio links.
What happened
AstroForge, a startup focused on asteroid mining, has developed an autonomous control system called Solo, a transformer-based AI model that will be onboard its upcoming spacecraft. The AI is designed to handle various critical spacecraft functions such as navigation, power management, and fault resolution without direct intervention from Earth-based operators. This marks a departure from traditional spacecraft operations, which typically rely on extensive human oversight and ground-based control centers.
The company plans to fly its first AI-controlled spacecraft mission, Autonomy-1, without traditional radio communications to Earth, effectively putting the AI in full command. Prior efforts with earlier prototype spacecraft faced communication challenges, motivating AstroForge to explore onboard AI autonomy. Solo will first undergo a trial in 'shadow mode' on the DeepSpace-2 mission, launching alongside Intuitive Machines’ third lunar mission, before assuming full control on Autonomy-1.
Why it matters
Space missions have traditionally depended on costly and complex ground infrastructure for mission control, requiring teams of operators and extensive communication networks. For startups like AstroForge, replicating this scale of support is financially prohibitive. By embedding AI autonomy on the spacecraft itself, the company aims to reduce operational costs and increase mission resilience against communication blackouts or unforeseen anomalies.
The use of a transformer-based AI model trained on thousands of spacecraft sensors represents a cutting-edge application of neural networks in spaceflight. While traditional spacecraft autonomy has relied mainly on classical control algorithms due to concerns over neural network reliability, AstroForge’s approach leverages advances in AI to potentially allow spacecraft to self-diagnose and recover from anomalies independently, transforming how deep space exploration and resource missions are conducted.
What to watch next
AstroForge’s next key milestone is the launch of its DeepSpace-2 spacecraft by the end of 2026. On this mission, the Solo AI will operate in shadow mode, where it runs alongside traditional systems without controlling the spacecraft, allowing engineers to evaluate its performance in real-time space conditions. This trial is crucial for validating Solo’s capabilities and reliability before full deployment.
Following this, the Autonomy-1 mission scheduled for 2027 will test the feasibility of spacecraft operating entirely under AI control, potentially without any Earth-based radio communication. Observers will be keen to see how effectively the AI can manage navigation, power, and unexpected issues autonomously, confirming whether AstroForge’s bold approach may set a new standard for small spacecraft and asteroid missions in the coming years.