GPS Denied IBVS-Based Navigation and Collision Avoidance of UAV Using a Low-Cost RGB Camera

2025-09-22

AI Summary

In GPS-denied environments, UAVs face significant challenges in achieving multi-target navigation and dynamic obstacle avoidance using only low-cost monocular RGB cameras. Method: This paper proposes a fully onboard image-based visual servoing (IBVS) framework that integrates AprilTag-based visual target detection, AI-driven monocular depth estimation, and a lightweight IBVS controller, implemented end-to-end in real time on a Jetson embedded platform—requiring no stereo cameras, external localization, or explicit path planning. Contribution/Results: The key innovation is the first deep integration of monocular depth estimation into the IBVS closed-loop, enabling a unified visual servoing strategy that simultaneously supports goal-directed navigation and obstacle perception. Experimental results demonstrate stable autonomous traversal across multiple visual landmarks without GPS, with real-time dynamic obstacle avoidance, an average positioning error of <0.15 m, and a control frequency of 30 Hz—substantially enhancing autonomy and deployment flexibility.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

This paper proposes an image-based visual servoing (IBVS) framework for UAV navigation and collision avoidance using only an RGB camera. While UAV navigation has been extensively studied, it remains challenging to apply IBVS in missions involving multiple visual targets and collision avoidance. The proposed method achieves navigation without explicit path planning, and collision avoidance is realized through AI-based monocular depth estimation from RGB images. Unlike approaches that rely on stereo cameras or external workstations, our framework runs fully onboard a Jetson platform, ensuring a self-contained and deployable system. Experimental results validate that the UAV can navigate across multiple AprilTags and avoid obstacles effectively in GPS-denied environments.

Problem

• UAV navigation using only RGB camera without GPS
• Collision avoidance through monocular depth estimation
• Onboard processing without external workstations or stereo cameras

Innovation

• IBVS navigation without explicit path planning
• AI-based monocular depth estimation for collision avoidance
• Fully onboard execution on Jetson platform