SUAS 2025 Development & Test

We established a data strategy by identifying key image sources, including open datasets, synthetic imagery, and drone footage. Initial image collection and camera hardware tests were conducted to ensure compatibility with our labeling system.

Collected images were labeled for model training. A modular preprocessing pipeline was developed using techniques such as Unsharp Masking and Gaussian filtering to improve input quality.

We set up the Gazebo simulation environment integrated with PX4 and MAVSDK. Initial mission scenarios were tested virtually to evaluate autonomous flight behavior and emergency protocols.

Training began using YOLOv8 with a fully prepared dataset. Data augmentation techniques were applied, and evaluation metrics like mAP and inference speed guided model tuning.

Simulation data informed key flight parameters like altitude and speed. We implemented feedback-based autopilot adjustments using real-time telemetry emulation to enhance system stability.

Python scripts were created to geotag drone images with GPS data. This enabled accurate image-to-map alignment for future mapping and 3D reconstruction tasks.

The trained model was deployed in live drone missions. Real-time object detection was tested, with a focus on latency, precision, and false positive rates.

We processed captured data in Pix4D to generate high-resolution 2D and 3D maps. Detection outputs and terrain models were analyzed to assess system performance.

A final report was prepared summarizing the technical results, highlighting strengths, and outlining areas for improvement in future development cycles.

We began designing the payload mechanism, focusing on balanced placement of four payloads. Initial 3D prints were taken, and the design was refined through iterations.

Drone dimensions were defined based on the payload system and propellers. Materials were selected with an emphasis on strength and lightness, shaping the main production plan.

Initial drone parts (motor mounts, ESC holders, landing gear) were modeled in SolidWorks and 3D printed for testing. Revisions were planned based on print results.

We revised and strengthened existing part designs, improving durability. Additional necessary mechanical components were added, enhancing the overall design quality.

All parts were finalized and assembled virtually in SolidWorks. The complete 3D model helped identify and resolve potential fit or compatibility issues.

Stress and force analyses were conducted using ANSYS. Weak areas in the design were identified, guiding modifications to improve structural durability.

Key parts were redesigned for better strength-to-weight ratio and reprinted. Carbon fiber components were prepared to reinforce the body, arms, and legs.

The drone was fully assembled. A support piece was added between the battery housing and legs, further strengthening the structure.

Weight balance and flight controller calibration were performed. Test flights led to small adjustments, finalizing the drone for competition readiness.

The team was formed and tasks were distributed. Necessary electronic systems for the project were identified. Research on motors, ESCs, and power distribution began.

Research continued. Circuit designs of similar systems were examined. Components to be used were finalized. Initial circuit designs were started.

Power board and motor driver circuit were designed. Suitable components were selected and circuit schematics were created. Board layouts were completed and made ready for production.

The first boards were produced. Soldering processes were completed. Basic electrical tests were started. Voltage and current measurements were conducted for verification.

Testing continued. ESC connections and power distribution were checked. Preparations were made for the integration of the boards into the flight system.

Field tests began. The operational stability of the electronic system was observed. A telemetry system was established, and initial data transmission trials were conducted.

Field testing continued. Errors identified during tests were resolved. Battery connections and power lines were reorganized. Overall board performance was improved.

Integration of the electronic system into the flight platform was completed. System stability was tested under different scenarios. Compatibility of all components was observed and results were recorded.

All checks were repeated. Boards and connections were reviewed. A technical report was prepared.