paper
Self-Refining Agentic Reinforcement Learning for Vision-Conditioned UAV Navigation
paperactiveprovisional
self-refining-agentic-reinforcement-learning-for-vision-conditioned-uav-navigation-3e99400b·1 events·first seen 14d agoAliases: Self-Refining Agentic Reinforcement Learning for Vision-Conditioned UAV Navigation
Co-occurring entities
More like this (12)
EvolveNav: Proactive Preflection and Self-Evolving Memory for Zero-Shot Object Goal NavigationVisual Verification Enables Inference-time Steering and Autonomous Policy ImprovementUniIntervene: Agentic Intervention for Efficient Real-World Reinforcement LearningEntropy Is Not Enough: Unlocking Effective Reinforcement Learning for Visual Reasoning via Vision-Anchored Token SelectionTempoVLA: Learning Speed-Controllable Vision-Language-Action PoliciesReinforcement Learning for CodeGoal-Conditioned Reinforcement Learningdecoupled reinforcement learningRLIF (Reinforcement Learning from Internal Feedback)Human-AI Teaming Through the Lens of CalibrationUsing Reward Uncertainty to Induce Diverse Behaviour in Reinforcement LearningEntropy-Regularized Reinforcement Learning
Recent events (1)
AgenticRL: Self-refining LLM-guided reward design and policy refinement for UAV navigation
AgenticRL is a framework that uses a multimodal GPT agent to automate reward function generation, policy training via PPO, and closed-loop self-refinement for UAV navigation tasks. The agent evaluates trained policies through diagnostic feedback, identifies failure modes, and iteratively refines rewards without human intervention. Evaluated across five navigation tasks, the closed-loop refinement improves policy behavior by 71% over initial rewards, with sim-to-real transfer achieving 91% real-world success rate and 94% sim-to-real accuracy.