Design

We conducted a literature review to inform our design development. This research indicated that a VTOL UAV would be the most suitable design for our intended mission. Our goal is to create a single platform that combines the advantages of both fixed-wing and rotary-wing UAVs—offering high-speed maneuverability while maintaining the ability to operate in confined spaces, hover for reconnaissance, and handle challenging terrain during takeoff and landing. This versatile approach will expand our mission capabilities and improve overall success rates.

Wing Design

We chose a high-wing configuration for the UAV design. This placement positions the center of mass below the center of lift, ensuring stable flight characteristics. The high-wing design also offers better protection for the wing profile from external damage during landing operations.

We chose a straight (rectangular) wing structure primarily for manufacturing simplicity. This design provides aerodynamic benefits—the larger air contact surface reduces speed loss during dives and improves gliding capability. The ailerons also help minimize wingtip turbulence.

Motors

We positioned tilt-rotors under the wings with movement capabilities along both x and y axes. This dual-axis movement ensures stability and ease of control during takeoff, landing, and flight operations. Tilt-rotors were selected for their ability to generate powerful thrust and lift forces—both independently and in combination. The rotors are configured to work in harmony with the fixed wings, preventing axial deviation during landing.

Electronic Equipment

We selected electronic equipment to maximize performance. The Pixhawk Cube Orange serves as our flight control board, chosen for its stable flight data, open-source architecture, and configurability for our UAV design. For precise positioning, we implemented the Pixhawk Here 3 CAN GPS. The RFD900x RF receiver offers an impressive range exceeding 40 km and data transfer speeds of 500 kb/s. The propulsion system consists of a T-Motor U7 V2.0 KV490 motor, T-Motor CF Series 16*5.4 propeller, Skywalker 80A ESC, and dual 6S 13000 mAh 40C Li-Po batteries—all selected to meet our thrust requirements. Safety features include a 175A fuse and circuit breaker. For video transmission, the SIYI HM30 FPV Combo delivers 1080p/60fps footage with a 30 km range and broad connectivity options. High-torque servos control motor rotation for thrust generation.

Software

We used QGroundControl as our primary ground station software. Additionally, our team designed and developed a custom interface as an alternative, using Python with the PyQt5 library. We configured our UAV's airframe through PX4 autopilot software and implemented an image transmission system to enable ground station image processing. The route from one of our autonomous flight tests are shown below.

Image Processing and Mapping

We used OpenCV for image processing and implemented the YOLO-V3 deep learning algorithm to ensure high speed and accuracy. The tent detection models were trained using Google Colab, and testing confirmed their successful creation. Screenshots below show both the training process and sample image detection results.

The mapping will be performed in two dimensions. The UAV's camera will capture high-resolution photos at set intervals. These photos will be processed using image processing and AI techniques to generate a map. The mapping process concludes by entering the target location's coordinates into the ground station and transferring them to the generated map.

Simulation

We use simulation primarily to test mission scenarios and evaluate algorithm efficiency without risking hardware. The system integrates ROS (Robot Operating System) with PX4 and the Gazebo simulator. PX4 interfaces with Gazebo to receive simulated sensor data and transmit motor commands. We incorporate detection target models into the simulation files. This environment allows us to test and optimize mapping and motion planning applications before real-world deployment.

Mission

The UAV performs autonomous takeoff and reconnaissance flights. Through image processing, it identifies friendly and enemy targets using intelligence data from the ground station. Upon receiving ground station commands, the UAV executes a dive maneuver toward the designated target while firing controlled bursts from its integrated airsoft weapon system.

A designated hangar location is marked in the mission area along with strategically placed targets (tents and human figures). After autonomous takeoff from the hangar, the UAV begins its reconnaissance mission. During flight, it creates a map of the area and uses image processing to identify and classify friendly and enemy targets, transmitting their coordinates to the ground station. Following ground station commands, the UAV either maintains its reconnaissance pattern or initiates a dive attack on specified coordinates before returning to the hangar.

Aircraft Specifications

• Weight : 8.5 kg
• Width : 2 m
• Flight Time : 30 min
• Cruise Speed : 17 m/s