Self-Refining Agentic Reinforcement Learning for Vision-Conditioned UAV Navigation
2026-06-02 • Robotics
RoboticsArtificial Intelligence
AI summaryⓘ
The authors developed AgenticRL, a system that helps drones learn to navigate by automatically creating and improving the rules (rewards) they use to learn, without needing humans to design these rules manually. Their system uses a smart language model to understand the task and environment, train the drone, evaluate its performance, and adjust the learning process in a loop. This system also helps select the right learned behavior for real-world situations by analyzing images and task descriptions. Their tests showed that this approach improved the drone's navigation skills significantly and worked well when transferring from simulation to real-world flying.
Deep reinforcement learningReward functionsProximal Policy Optimization (PPO)Multimodal GPT agentUnmanned Aerial Vehicles (UAV)Policy refinementSim-to-real transferObstacle avoidanceTrajectory following
Authors
Roohan Ahmed Khan, Yasheerah Yaqoot, Muhammad Ahsan Mustafa, Dzmitry Tsetserukou
Abstract
Deep reinforcement learning has shown strong potential for enabling autonomous robots to learn complex navigational tasks. However, its practical use still depends heavily on human designed reward functions and repeated manual fine tuning, which is time consuming and does not guarantee high success in the desired task. This paper presents AgenticRL, agent guided reinforcement learning framework that increases autonomy in reward design, policy refinement, and real world deployment for unmanned aerial vehicles (UAV) navigation tasks. AgenticRL uses a multimodal generative pre-trained tansformer (GPT) agent to interpret task information and visual scene observations, generate task specific reward functions, train policies using Proximal Policy Optimization (PPO) algorithm, and then act as a critic by evaluating the trained policy through diagnosis packets to generate feedback. Based on this feedback, the agent identifies failure modes and refines the reward function in a closed loop self improvement process. To further leverage the multimodal GPT agent during inference, AgenticRL uses real world images and natural language task information to automatically identify the active scenario and select the appropriate trained policy for execution. The framework is evaluated on multiple navigational tasks, including gate traversal, obstacle avoidance, wall barrier crossing with landing, trajectory following, and motion behavior learning. Experimental results show that the closed loop refinement process improves policy behavior compared with initial rewards by 71%. We also demonstrate sim-to-real transfer of the proposed framework, achieving a real world success rate of 91% and a sim-to-real accuracy of 94%.