LeLaR: The First In-Orbit Demonstration of an AI-Based Satellite Attitude Controller

📰 ArXiv cs.AI

LeLaR demonstrates the first in-orbit use of an AI-based satellite attitude controller using Deep Reinforcement Learning (DRL)

advanced Published 1 Apr 2026

Action Steps

Design and train a DRL agent in a simulation environment
Deploy the trained agent on a satellite and test its performance in orbit
Overcome the Sim2Real gap by adapting the agent to real-world conditions and uncertainties
Refine the agent's control strategies through continuous learning and feedback

Who Needs to Know This

Space engineers and AI researchers can benefit from this technology, as it enables more efficient and adaptive satellite control systems

Key Insight

💡 DRL can learn adaptive control strategies for satellite attitude control, overcoming limitations of classical controllers