Development and Testing for Perception-Based Autonomous Landing of a Long-Range QuadPlane

QuadPlanes combine the range efficiency of fixed-wing aircraft with the maneuverability of multi-rotor platforms, making them well-suited for long-range autonomous missions. In GPS-denied or cluttered urban environments, perception-based landing becomes essential for ensuring reliable operation. Unlike structured landing zones, real-world sites are often unstructured and highly variable, requiring strong generalization capabilities from the perception system. Deep neural networks (DNNs) offer a scalable solution for learning landing site features across diverse visual and environmental conditions. Although perception-driven landing has been successfully demonstrated in simulation, real-world deployment introduces significant challenges. Payload and volume constraints limit the integration of high-performance edge AI devices such as the NVIDIA Jetson Orin Nano Super, which are crucial for real-time object detection and control. Alongside perception-based control, accurate pose estimation during descent is necessary, especially in the absence of GPS, and relies on dependable visual-inertial odometry. Achieving this with the limited computational resources available on edge AI devices requires careful optimization of the entire deployment framework. The flight characteristics of large QuadPlanes further complicate the problem. These aircraft exhibit high inertia, reduced thrust vectoring capability, and slower response times, all of which contribute to the complexity of achieving precise and stable landing maneuvers. This work presents the design, integration, and field testing of a lightweight QuadPlane framework capable of efficient vision-based autonomous landing and visual-inertial odometry, specifically developed for long-range QuadPlane operations such as aerial monitoring. It describes the hardware platform, sensor configuration, and embedded computing architecture designed to meet demanding real-time and physical constraints. This system establishes the foundation for future deployment of autonomous landing technologies in dynamic, unstructured, and GPS-denied environments.