Recent advances in computation techniques and the latest progress in computing hardware hold the promise to transform aerospace guidance and control technologies and to dramatically increase the level of autonomy of future aerospace vehicles. Traditional designs, based on analytical solutions, can be replaced by algorithms that, by intensively relying on onboard computation, can accomplish the complex guidance and control tasks associated with autonomous operations. Model predictive control (MPC) falls into this class of methods and it is recognized by the aerospace community as one of the most promising ones, due to its systematic treatment of constraints, optimized performance, and robustness to uncertainty.
This research aims at designing novel guidance algorithms that embed convex optimization into the MPC framework to exploit the computation speed and deterministic guarantees of convergence of convex programming. Since most real-world aerospace problems are not naturally convex, state-of-the-art convexification methods are examined to make general optimization problems computationally tractable, thus enabling their real-time implementation. Furthermore, the research seeks to analyze the performance and accuracy properties of several discretization methods and evaluate their suitability for time-critical applications.
The primary contribution of this research is the application of state-of-the-art convexification methods to new classes of aerospace problems, never tackled in this way in the available literature, in order to embed the convex optimization problem into an MPC framework. This enables solving in real-time an OCP that accounts for the original problem dynamics and constraints, thus paving the way for future autonomous guidance algorithms.
Besides, this research will also provide an extensive and fair comparison of the computational performance and solution accuracy of several popular discretization schemes. Indeed, multiple discretization methods will be examined on several aerospace problems for which such a detailed analysis is not yet available. Particular attention will be paid on assessing their potentiality for integration in the convex MPC architecture.
Finally, by investigating the particular problem of VEGA's third stage guidance, the research has the final aim to outline a novel and more efficient method, compared to the current neutral-axis maneuver, to control the impact point of the spent stage while guaranteeing high performance. Although being a specific application, the general approach will be straightforwardly extendable to other classes of aerospace problems, thus providing a general contribution to the ongoing research in autonomous guidance.