Paragon Transwing AI-UAV
Abstract
The Paragon Transwing is an autonomous “Vertical Take-Off & Landing” aircraft which utilizes a full-swing wing transition mechanism to transition between Fixed-Wing and Quad-Rotor flight modes. The system is equipped with an NVIDIA Jetson Nano and has five primary subsystems; Flight Control, Power Management, Propulsion, Image Recognition, and Ground Control. Utilizing the Nano, the image recognition system is built upon the neural net framework “Darknet”. This process utilizes multiple scripts to manage object detection, identification, logging, and mission upload. A Mission Viewer dashboard was constructed using IBM Cloud, providing global remote-access to the system, and allowing wireless upload of mission and image data via an internet connection. There are three main components to the cloud solution; the web app, the Cloudant database, and Object Storage. The control system utilizes a Pixhawk Cube Black flight controller equipped with Real-Time Kinematic GPS, air data sensors, and mission planning files to autonomously execute a variety of missions. Mission files and aircraft telemetry are uploaded and monitored through the ArduPilot Mission Planner and the Ground Control Station. Potential applications of this technology include defense, emergency response, aerial surveying, and infrastructure analysis. This project is the flight system for the ME Project: Electric Vertical Take-Off and Landing (eVTOL)
