Open Source Tools for Deployment of GPS-Denied Autonomous UAVs in Real-World Applications

Anonymous

Keywords: GPS-denied, UAV, Semantic SLAM, Drift Correction, Planning Under Uncertainty, Map inflation

TL;DR: We present an open-source autonomy stack for UAVs operating in GPS-denied environments, focusing on improving state estimation accuracy and uncertainty aware trajectory planning.

Abstract: Despite the widespread use of Autonomous Unmanned Aerial Vehicles (UAVs) in various outdoor applications, their deployment in indoor and GPS-denied environments—such as under dense forest canopies—continues to pose significant challenges. Robust trajectory planning, accurate state estimation, and effective obstacle avoidance are crucial to successful mission completion in these complex, dynamic, and unstructured settings. In this work, we present our open source autonomy stack for aerial robots operating in GPS-denied environments. We specifically focus on two critical components: (1) compensation for inaccurate state estimation and (2) trajectory planning in uncertain scenarios. Using these techniques, our systems are capable of executing long-range missions in challenging environments. We hope that the tools and methodologies introduced in this study will accelerate the adoption and safe deployment of UAVs in GPS-denied conditions.