A Geometric Approach to Trajectory Planning for Underactuated Mechanical Systems

In the last decade, multi-rotors flying robots had a rapid development in industry and hobbyist communities thanks to the affordable cost and their availability of parts and components. The high number of degrees of freedom and the challenging dynamics of multi-rotors gave rise to new research problems. In particular, we are interested in the development of technologies for an autonomous fly that would al- low using multi-rotors systems to be used in contexts where the presence of humans is denied, for example in post-disaster areas or during search-and-rescue operations. Multi-rotors are an example of a larger category of robots, called “under-actuated mechanical systems” (UMS) where the number of actuated degrees of freedom (DoF) is less than the number of available DoF. This thesis applies methods com- ing from geometric mechanics to study the under-actuation problem and proposes a novel method, based on the Hamiltonian formalism, to plan a feasible trajectory for UMS. We first show the application of a method called “Variational Constrained System approach” to a cart-pole example. We discovered that it is not possible to extend this method to generic UMS because it is valid only for a sub-class of UMS, called “super-articulated” mechanical system. To overcome this limitation, we wrote the Hamilton equations of the quad- rotor and we apply a numerical “di- rect method” to compute a feasible trajectory that satisfies system and endpoint constraints. We found that by including the system energy in the multi-rotor states, we are able to compute maneuvers that cannot be planned with other methods and that overcome the under-actuation constraints. To demonstrate the benefit of the method developed, we built a custom quad- rotor and an experimental setup with different obstacles, such as a gap in a wall and we show the correctness of the trajectory computed with the new method.