Artificial intelligence for unmanned aerial systems

Modern unmanned aerial systems (UAS) face a significant challenge: limited autonomy. Many vehicles still depend on human operators, which can restrict their effectiveness and increase the likelihood of errors. Integrating artificial intelligence directly into onboard control systems can help address this challenge.

NNTrack logo

Brain Development LLC and Robotrack LLC develop and manufacture educational systems for children and young people. They have developed NNTrack, a visual programming environment for designing neural-network architectures and training custom networks with accessible block-based tools.

Neural-network software interface

An updated version of the software supports YOLO (You Only Look Once) neural networks and enables trained models to run on unmanned aerial vehicles (UAVs). This adds image-processing and real-time environment-analysis capabilities, which are especially useful for UAV navigation and situational awareness.

Example applications

  1. Agricultural drones: identifying diseased plants and weeds and estimating crop yields.
  2. Industrial and urban drones: finding defects in pipelines, power lines, and building roofs; assessing road surfaces; and identifying road incidents.
  3. Search drones: detecting people, animals, or vehicles in challenging conditions.

Ways to integrate neural networks into UAS

  • Run trained models directly on the drone. The onboard controller must have enough processing power for video, or the UAV can carry an additional single-board computer, such as a Raspberry Pi 5 with 8 GB of memory and a webcam.
  • Stream video from the UAV to NNTrack over Wi-Fi for analysis on a computer, then send the neural network’s processed results back to the UAV.

Clover quadcopter

Integration with the Clover UAV has already been implemented. Clover is an educational programmable quadcopter kit built from popular open components and supplied with documentation and software libraries. The solution was used in the forest fire safety track at Archipelago 2025, an annual project and education intensive held as part of the International Forum for the Development of Unmanned Systems.

Invitation to collaborate

Drone manufacturers are invited to integrate their products with NNTrack. For more information or a consultation, contact the team at +7 (921) 330-25-68 or mrtrus2014@yandex.ru.