Lorenzzo Mantovani
Reinforcement learning for autonomous spacecraft decision-making
I leverage deep reinforcement learning and planning methods to enable autonomous decision-making in complex and uncertain environments.
My research lies at the intersection of aerospace engineering and machine learning, with the goal of enabling reliable autonomy in safety-critical systems such as spacecraft and multi-agent space missions.
GitHub · LinkedIn · Google Scholar ·
Research Focus
Autonomous Earth-observing spacecraft: reinforcement learning for real-time satellite tasking under uncertainty, including cloud coverage and long-horizon decision-making.
Safe and robust learning-based control: integrating safety mechanisms such as shielding and training strategies like curriculum learning to ensure reliable operation of autonomous spacecraft.
Distributed and adversarial autonomy: multi-agent coordination for satellite constellations and reinforcement learning approaches for adversarial agent scenarios.
Featured Highlight
Capture the Satellite Challenge — AIAA SciTech (2026)
Our CU Boulder team placed Second Place in the Capture the Satellite Challenge, developing autonomous spacecraft control strategies for an on-orbit interception scenario.
Selected Projects
Earth-Observing Satellites
Reinforcement learning for agile Earth-observing spacecraft, including shielding, curriculum learning, and cloud coverage uncertainty.
Multi-Agent Systems
Distributed and intent-sharing approaches for autonomous satellite constellations and large-scale coordination.
Adversarial Autonomy
Decision-making in adversarial spacecraft pursuit-evasion scenarios.