Developers presented a drone flight simulator based on the connectome of a fly

The FlyDrones project offers a browser-based simulation of 3D drone control, operating on a biologically accurate model of the fruit fly's nervous system. The system simulates the work of 166,000 neurons of the insect's connectome to process visual data in real-time.

The algorithm uses mechanisms of optical flow and basic reflexes characteristic of biological organisms to navigate in virtual space. Users can control the drone's flight using hand gestures or through a webcam, which transmits movements into commands for the simulator's neural network.

The project's code is published on the GitHub platform under an open license. The software implementation allows observing how impulses are transmitted within the simulated neural structure, ensuring stabilization and obstacle avoidance without using classical computer vision algorithms.

This development demonstrates the potential of applying neurobiological models in robotics. Using an insect connectome allows achieving high autonomy with minimal computational costs, opening up prospects for creating compact control systems for unmanned vehicles capable of functioning in conditions of limited resources.