A bio-motivated vision system and artificial neural network for autonomous UAV obstacle avoidance

Unmanned aerial vehicles (UAVs) becoming more and more common. They show excellent potential for multiple types of autonomous work, although they must achieve these tasks safely. For flight-safety, it must be assured that the UAV will avoid collision with any objects in its flight path during autonomous operations. Computer vision and artificial neural networks have shown to be effective in many applications. However, biological vision systems and the brain areas responsible for visual processing may hold solutions capable of acquiring information effectively. We are proposing a novel system, which performs visual cue extraction with algorithms based on the structure and functionality of the retina and the visual cortex of the mammalian visual system, and a convolutional neural network processing data to detect a predefined obstacle using the onboard camera of the UAV. We also examined the effect of preprocessing on calculation time and recognition effectiveness.