Reliability-Asymmetric Spacecraft Autonomy: Co-Designing a Capable Learned GNC Stack with a Verified, Adaptation-Aware Runtime Shield
📰 ArXiv cs.AI
Learn how to co-design a reliable spacecraft autonomy system using a learned GNC stack with a verified runtime shield, enabling both capability and certifiability
Action Steps
- Design a three-tier GNC stack using a foundation-model commander, a constraint-screening verifier, and a fault-adaptive controller
- Implement a capability path that combines natural language mapping to PDDL+ with bounded verification
- Develop a verified, adaptation-aware runtime shield to ensure reliability and certifiability
- Test and validate the AMPLE-GNC system using simulation-based evaluation
- Integrate the learned GNC stack with the runtime shield to enable reliable spacecraft autonomy
Who Needs to Know This
This research benefits aerospace engineers, AI researchers, and spacecraft autonomy specialists who need to develop reliable and certifiable onboard autonomy systems for deep-space missions
Key Insight
💡 Co-designing a capable learned GNC stack with a verified runtime shield can achieve both reliability and certifiability for deep-space missions
Share This
🚀 Enable reliable spacecraft autonomy with AMPLE-GNC, a co-designed learned GNC stack and verified runtime shield #spacecraftautonomy #AI
Key Takeaways
Learn how to co-design a reliable spacecraft autonomy system using a learned GNC stack with a verified runtime shield, enabling both capability and certifiability
Full Article
Title: Reliability-Asymmetric Spacecraft Autonomy: Co-Designing a Capable Learned GNC Stack with a Verified, Adaptation-Aware Runtime Shield
Abstract:
arXiv:2606.25366v1 Announce Type: cross Abstract: Deep-space missions need onboard autonomy that is both capable and certifiable. Rule-based autonomy is certifiable but brittle, while learned autonomy is capable but hard to verify. We present AMPLE-GNC, a three-tier guidance, navigation, and control stack. Its capability path combines a small foundation-model commander that maps natural language to PDDL+, a constraint-screening verifier, and a fault-adaptive controller. All three are bounded by
Abstract:
arXiv:2606.25366v1 Announce Type: cross Abstract: Deep-space missions need onboard autonomy that is both capable and certifiable. Rule-based autonomy is certifiable but brittle, while learned autonomy is capable but hard to verify. We present AMPLE-GNC, a three-tier guidance, navigation, and control stack. Its capability path combines a small foundation-model commander that maps natural language to PDDL+, a constraint-screening verifier, and a fault-adaptive controller. All three are bounded by
DeepCamp AI