Robust “blind” navigation for a miniature ducted-fan aerial robot

This work presents a control strategy to allow a miniature ducted-fan aerial robot to perform flight operations in an unknown environment potentially cluttered with obstacles. As a main challenge, it is assumed that the vehicle is completely “blind”, namely that obstacles cannot be sensed a priori so as to be avoided or even detected in real-time by means of contact or force sensors installed on the vehicle. As a consequence, the control design has to be robust with respect to the presence of possible unexpected contacts with the surrounding environment. In order to maintain stability in such a complex scenario, the proposed approach relies both on some mechanical properties of the aircraft layout, such as in particular the relative position of possible contact points with respect to the onboard actuators, and on the presence of a supervisor able to detect the presence of an obstacle only by observing the behavior of the flight control loop. The effectiveness of the obtained results is then demonstrated by means of experiments on a real setup.