CALOS: Control-Affine Lyapunov On-manifold Safety Layer for Safe Deep Reinforcement Learning for Quadrotors
Bridging the gap between high-performance deep reinforcement learning and formal safety guarantees in aerial robotics.
Deep Reinforcement Learning has demonstrated remarkable capability in quadrotor control, yet learned policies offer no guarantee of respecting safety constraints during training or deployment. We present CALOS (Control-Affine Lyapunov On-manifold Safety), a runtime safety layer t...
ArXiv CS.RO
Source
Generated insight by OnAirToday AI
CALOS bridges the gap between high-performance neural network policies and formal safety guarantees by enforcing physical constraints through Lyapunov-based manifold control. This approach marks a critical shift toward deploying autonomous drones in complex environments where failure is not an option.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.