Exp2VLA: Enabling Vision-Language-Action for Drone Navigation from Expert Demonstrations
2026-07-03 • Robotics
Robotics
AI summaryⓘ
The authors focus on making drone control easier by combining vision, language, and action in one model. They create a method called Exp2VLA that learns from expert behaviors, like those from advanced controllers or human operation, to teach smaller, efficient models how to navigate using language instructions. Their approach helps drones understand complex commands and work well even with new challenges. This makes it simpler to develop flexible drone behaviors without complex setups.
vision-language-action modelsdrone navigationexpert distillationreinforcement learningteleoperationlanguage-conditioned controlsim-to-simmechatronic systemsrobot intelligencesemantic commands
Authors
Van Huyen Dang, Kabilesh Rajendran, Erdi Sayar, Erdal Kayacan
Abstract
Vision-language-action (VLA) models open a new path toward intuitive robot control by directly linking perception, language, and action in a single end-to-end framework. Yet for UAVs, practical adoption remains difficult because existing solutions are either computationally heavy or insufficiently capable in complex environments. In this work, we propose a practical expert-distillation pipeline (Exp2VLA) for language-conditioned drone navigation. The core idea is to distill expert behavior, obtained from reinforcement learning, teleoperation, or other controllers, into training data that can be used to fine-tune compact VLA models. This allows existing control strategies to be transferred into a unified language-guided navigation model, reducing manual system integration and lowering the barrier for deploying new robot behaviors. Experiments in both sim-to-sim and simulation-in-the-loop settings across multi-object scenes show that the fine-tuned models can handle varied semantic commands and generalize to unseen target compositions. The proposed framework demonstrates how expert-policy distillation can help mechatronic systems move from specialized control modules toward more flexible and reusable robot intelligence.