The Optimization Process of an Autonomous UAV

Votes: 2
Views: 4199

Paradoxically, the idea that underlies “On the Origin of Species,” written by Charles Darwin comes from geology. Just as rivers have created large geological formations so natural selection has created animals with an optimal configuration, relative to the environment in which they live. Recent research has uncovered that some features of the fins and of the tail of cetaceans would appear to contradict some aerodynamic theories. Bionicists are convinced of the performance of nature’s way of optimization. This sophisticated method is running already 3 billion years, in a really big laboratory: the surface of the Earth. Now, the laboratory is large enough and the time is long enough to find good solutions (perhaps optimal solutions) under the biological constraints.

The idea of using this type of mechanism, writing an appropriate genetic algorithm, can make it much easier to choose the optimal configurations between all those that are created in the preliminary phase of the project of a UAV. In addition to properly characterizing the genotype of the aircraft, the mechanisms of crossover and mutation, is essential the definition of a correct fitness function. This function measures the adaptation to the environment of the configurations produced. It is not possible, for obvious reasons, to replicate the airfoils of the wing of a bird, so the fitness function must contain more terms. Each term represents a technological constraint or an environmental constraint. The fitness function contains a term related to the purpose of the UAV, e.g. a scout with a high time range, structural constraints, the wing of a bird is optimal but too thin and does not respect the constraints of rigidity required for an aircraft, and stability requirements. Stability is essential, the design of the wing affects the design of the tail.

The aerodynamic design of the aircraft also includes the choice of moving surfaces. The UAV will have a complex on-board electronics, an autopilot and a variety of sensors and actuators. The subsumption architecture, with its layered structure, allows to build a new type of avionics based on a revolutionary model, for the field of aeronautics, of the relationship between perception and action. The use of this architecture allows to build an artificial intelligence that controls all the aspects of a mission harmonizing them. The subsumption architecture is based on the idea that intelligence is an emergent property of complex systems. Then a suitable choice of sensors and actuators for each level that characterizes the architecture allows to prepare the way for the execution of complex tasks, or missions. The reactive model, at the base of subsumption, allows quick responses to unexpected or urgent events. The development of an artificial intelligence based on this type of architecture would prepare the way for the development of abstract thinking, which would allow for example to identify objects.

By now I obtained good results with the optimization process of airfoils. I'm working on the definition of the fitness function, on the terms related to each constraint. I published an article related to subsumption theory.


Voting is closed!


  • Name:
    Filippo Menconi
  • Type of entry:
  • Profession:
  • Filippo's favorite design and analysis tools:
    Catia, Matlab, SimMechanics, Simulink
  • Filippo's hobbies and activities:
    autonomous UAVs, robotics, cognitive sciences
  • Filippo belongs to these online communities:
    Twitter, Researchgate, Google+, Mendeley
  • Filippo is inspired by:
    I try to surpass the standard way of thinking, I try to connect different sciences, or different fields of engineering. I always try to find unconventional solutions without loosing the advantages of the conventional ones.
  • Software used for this entry:
    Xflr5 Xfoil Matlab Python
  • Patent status: