A DRL policy learns racing controls from depth spectral distributions using a non-geometric physics-informed reward, achieving 12% better performance than humans on out-of-distribution tracks with under 1% of baseline computation.
Champion-level drone racing using deep reinforcement learning
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
fields
cs.RO 4years
2026 4verdicts
UNVERDICTED 4representative citing papers
MARS-Dragonfly creates a force-torque virtual quadrotor model and two-stage predictive allocator that lets reconfigurable drone modules fly stably and agilely, validated in real-world experiments with 0.0896 m average position error.
GaussFly decouples representation learning from policy optimization via 3D Gaussian Splatting reconstruction and contrastive features to achieve superior sample efficiency and zero-shot sim-to-real transfer for AAV visuomotor policies.
A universal LLM-to-drone interface is implemented via the Model Context Protocol (MCP) and Mavlink, demonstrated with real UAV flight control and simulated flights using live map data.
citing papers explorer
-
Physics-Informed Reinforcement Learning of Spatial Density Velocity Potentials for Map-Free Racing
A DRL policy learns racing controls from depth spectral distributions using a non-geometric physics-informed reward, achieving 12% better performance than humans on out-of-distribution tracks with under 1% of baseline computation.
-
MARS-Dragonfly: Agile and Robust Flight Control of Modular Aerial Robot Systems
MARS-Dragonfly creates a force-torque virtual quadrotor model and two-stage predictive allocator that lets reconfigurable drone modules fly stably and agilely, validated in real-world experiments with 0.0896 m average position error.
-
GaussFly: Contrastive Reinforcement Learning for Visuomotor Policies in 3D Gaussian Fields
GaussFly decouples representation learning from policy optimization via 3D Gaussian Splatting reconstruction and contrastive features to achieve superior sample efficiency and zero-shot sim-to-real transfer for AAV visuomotor policies.
-
A Universal Large Language Model -- Drone Command and Control Interface
A universal LLM-to-drone interface is implemented via the Model Context Protocol (MCP) and Mavlink, demonstrated with real UAV flight control and simulated flights using live map data.