Back to main page / Back to main page

Jump to the page in German language

Drones with Artificial Intelligence - Master Project - Study Field Intelligent Systems (WS2526)

Participants in the project form groups of up to four people. Each group is provided with an AI-capable FPV drone and the necessary components. Their tasks are the development, implementation, investigation, and evaluation of exciting and comprehensible drone AI applications.

• Task 1: Become familiar with the FPV drone and the additional hardware components (single-board computer, camera, A/V video grabber, AI accelerator, etc.).
• Learn about the capabilities and limitations of the hardware components. These include:
• Frame, flight controller and electronic speed controller, FPV transmitter (for video), ELRS receiver, motors, GPS receiver, camera(s), batteries, remote control, single-board computer (e.g. Raspberry Pi Zero 2 WH), AI accelerator (e.g. Google Coral USB Accelerator), FPV goggles, A/V video grabber, etc.
• Learn about the capabilities and limitations of the software components. These include:
• Flight controller firmware (Betaflight, INAV, ArduPilot), ground control station (e.g. QGroundControl for ArduPilot), operating system (e.g. Raspberry Pi OS, Ubuntu) for the single-board computer, AI software (e.g. TensorFlow Lite, YOLO), etc.
• The focus of the project is on the AI applications (see Task 2)! Building a new drone is neither necessary nor the goal of the project. Modifications to the existing drones are possible if desired.
• Task 2: Develop, integrate, and test AI application(s). Possible application areas include:
• Computer Science: Object recognition (e.g. detection of people and animals)
• Logistics: Automated delivery and pickup of objects
• Geography: Biodiversity research or detection of storm damage
• Engineering: Structural inspections
• Agricultural Sciences: Monitoring of plants and pest infestation
• Archaeology: Aerial imaging
• Atmospheric Research: Smog measurements
• Task 3: Research and integrate autopilot options.
• For example using INAV and ArduPilot.
• Task 4: Research and integrate delivery/payload options.
• Develop and test a simple drop mechanism (e.g. using a servo).
• Task 5: Documentation and presentation of the results of Tasks 1–4.
• You will develop documentation and instructions that enable students, researchers, and lecturers to reproduce the AI drone scenarios and use them for their own modules and research projects.
• There will be no slide presentations or PDF project reports! Each team develops a complete and comprehensible online documentation (e.g. via GitHub Pages) and presents their results with a poster and a live demonstration.

Schedule for the Semester

Documents

		AI Drones: A bilingual handbook on the design, construction, and use of AI-capable drones for scientific projects and teaching (version: 16.11.2025)
		Project description
		Project description

Equipment

Each team receives the following equipment:

• 3.5'' FPV drone with CineWhoop frame (including propeller guards), fully assembled and ready to fly
• Skyzone Cobra X FPV goggles
• Radiomaster Boxer ELRS transmitter (remote controller) with EdgeTX firmware v2.11.3 and binding phrase configured to match the receiver: drone[1-3]
• Google Coral USB Accelerator AI accelerator
• Raspberry Pi Zero 2 WH single-board computer
• A Raspberry Pi OS Lite (32-bit) based on Debian 11.11 (Bullseye) is installed on the SD card. With this version, the libraries required for the Google Coral USB Accelerator and the necessary Python version can be installed very easily.
• Raspberry Pi Camera Module v2
• Mini-HDMI to standard HDMI cable
• Micro-USB (m) to USB Type A (f) cable
• USB audio/video grabber (MacroSilicon MS210x)
• Smoke stopper (short-circuit protection plug)
• SkyRC B6neo+ charger
• Allen and hex keys in the sizes commonly used for FPV drones
• Spare propellers (3.5'')
• USB card reader for SD and microSD memory cards

The prepared drones are described in detail below...

Drone 1

• Frame: GEPRC Cinelog 35 GEP-CL35 3.5-inch CineWhoop Frame Kit
• Flight controller: Flywoo GOKU GN745 (STM32F745, 216 MHz, 1 MB Flash) 45A AIO 2–6S AM32 (Betaflight v4.5.2)
• Camera: Caddx Ratel Pro 1500TVL Analog
• Video transmitter: SpeedyBee TX800 VTX
• Channel: 5733 MHz (Band: BOSCAM/RichWave, Channel: 1)
• VTX antenna: Flywoo 60 mm Osprey 5.8 GHz RHCP SMA
• Receiver: Radiomaster XR1 ELRS Dual Band RX (Firmware: ExpressLRS 3.6.0)
• Binding phrase: drone1
• Motors: Emax Eco II 2004 3–6S 3000 KV
• Propellers: Gemfan 90 mm D90-5 3.5" ducted 5-blade propellers
• GPS: HGLRC M100 with compass

Drone 2

• Frame: SpeedyBee BEE35 Pro 3.5 CineWhoop Frame Kit
• Flight controller: Flywoo GOKU F722 PRO Mini V2 (STM32F722, 216 MHz, 512 kB Flash) 45A Stack 3–6S AM32 (Betaflight v2025.12.0-BETA)
• Camera: RunCam Phoenix 1000TVL Analog
• Video transmitter: SpeedyBee TX800 VTX
• Channel: 5771 MHz (Band: BOSCAM/RichWave, Channel: 3)
• VTX antenna: TrueRC Singularity 5.8 GHz RHCP SMA
• Receiver: Radiomaster XR1 ELRS Dual Band RX (Firmware: ExpressLRS 3.6.0)
• Binding phrase: drone2
• Motors: Emax Eco II 2004 3–6S 3000 KV
• Propellers: Gemfan 90 mm D90-5 3.5" ducted 5-blade propellers
• GPS: HGLRC M100 with compass

Drone 3

• Frame: SpeedyBee BEE35 Pro 3.5 CineWhoop Frame Kit
• Flight controller: Flywoo GOKU F722 PRO V2 (STM32F722, 216 MHz, 512 kB Flash) 55A Stack 3–6S AM32 (Betaflight v4.5.2)
• Camera: Caddx Ratel Pro 1500TVL Analog
• Video transmitter: SpeedyBee TX800 VTX
• Channel: 5847 MHz (Band: BOSCAM/RichWave, Channel: 7)
• VTX antenna: Foxeer Lollipop 4 RHCP
• Receiver: Radiomaster XR1 ELRS Dual Band RX (Firmware: ExpressLRS 3.6.0)
• Binding phrase: drone3
• Motors: Axisflying C206 2006 2500 KV 4–6S
• Propellers: Gemfan 90 mm D90-5 3.5" ducted 5-blade propellers
• GPS: Matek M10Q-5883 GPS with compass

Teams

Group 1

• Poster
• Documentation
• Project Page

Group 2

• Poster
• Documentation
• Project Page

Group 3

• Poster
• Documentation
• Project Page

Contact

You can best reach me via email: christianbaun@fb2.fra-uas.de