In this research authors propose the improved algorithm for collision prevention based on the combination of immune neural network and fuzzy logic for the team of autonomous unmanned electrical aerial vehicles (UAVs) cooperatively reaching the common goal. The mathematical model for cooperative and safe task performance of the UAVs is developed, target function for self-organized learning of the immune neuro-fuzzy control system is defined. Information, decisions and corresponding actions of each team participant are distributed between UAVs, so each autonomous device should make and optimal decision between safety and performance criteria, i.e. for safe maneuver leading towards the team goal achievement. UAVs’ team members share the data from each other and learn to make better decisions by unsupervised learning immune neuro-fuzzy algorithm. The experimental proofs of mathematical and computer modelling, as well as practical experiments of the multi-rotor helicopter prototype are provided.