Coordination and control of fixed-wing aerial robots

Abstract

Multi-agent systems have received much attention in the last years, since they are able to perform tasks better, faster or even those that are unfeasible for single agents. This work presents a MILP (mixed integer linear programming) based formulation for the coordination of multiple robots that must follow closed intersecting paths persistently. The strategy consists of an offline planning for the robots' velocities, which prevents the need of online collision avoidance maneuvers. The robot model considers minimum and maximum speed constraints, which allows the strategy to be applied to fixed-wing aerial robots. Issues that are often disregarded in the literature of multi-robot systems are addressed in the present work, they are: minimum spatial separation; acceleration limits; and uncertainties on the speeds and positions. In order to control each robot on its specified path, a guidance vector field approach technique was developed. In this stage of the dissertation a better representation of the dynamics of an airplane is considered. The used reference model is non holonomic and has constrained input controls. A consolidated vector field strategy is used to guide the computation of the control inputs. Asymptotic stability is proven with Lyapunov Theory and ultimate bounds are found when actuation disturbances are taken into account. The developed control strategy also has advantages over the strategies in related works. Its main characteristic is the ability to deal with generic curve shapes in 3 dimensions. Simulations with up to 48 robots show the computational efficiency of the MILP based coordination strategy. The efficiency of the proposed guidance vector field controller is demonstrated with simulations with a 6 degrees of freedom, 12 states realistic aircraft model. A real experiment with 3 actual e-puck robots is presented in order to validate the multi-robot coordination and control strategy in a real world scenario.