Flying among obstacles

Abstract

Addresses the problem of the reactive behaviors of flying 3D robots evolving in dynamic and unknown environments. Collision avoidance and target following have been studied with a common frame based on a nonlinear control algorithm. This method is based on the deformable virtual zone approach, initially developed for land vehicles and submarines. The main characteristics of this approach is the fact that there is no geometrical modelling of the obstacles but a vectorial representation of the interaction between the robot and its environment. This approach allows one to implement fast control laws and can be seen as a efficient low-level algorithm for controlling motions both to avoid obstacles and to reach moving goals. Several experimental results in simulation, demonstrating a 3D flying robot, are proposed, and the implementation of the algorithm on a real small-size helicopter is discussed.