1,721,010 research outputs found
Indirect Optimization of Bang-Bang Control Problems and Applications to Formation Flying Missions
Full text linkThis thesis is focused on indirect optimization methods for the design of space missions, and, in particular, to a specific class of optimal control problems whose solution exhibits a discontinuous control law: the so called bang-bang optimal control.
Any attempt to solving such problems by using an indirect method without any specific treatment of the bang-bang control inevitably results into a failure, except for trivial problems.
The thesis compares two techniques, conceptually quite different, that aim to handle (or just to reduce) issues related to the discontinuous profile of the optimal control: the Multi-Bound Approach and the Continuation-Smoothing Technique.
These two approaches are first tried out/tested on a very simple case (the rocket-sled problem) and then applied to obtain the solution of two rather complex problems: the cooperative rendezvous and the deployment of a two-spacecraft formation that flies in a High Eccentricity Orbit (referring to the Simbol-X project).
The general philosophy that stands behind either approach is outlined, as well as relative strength and weakness. Range of applicability, effort required to the user, computational time, and convergence radius are analyzed and discussed.This work was supported by the Centre National d’Etudes Spatiales (CNES), [Contract Number 93333/00]
Robust interplanetary trajectory design under multiple uncertainties via meta-reinforcement learning
No full textThis paper focuses on the application of meta-reinforcement learning to the robust design of low-thrust interplanetary trajectories in the presence of multiple uncertainties. A closed-loop control policy is used to optimally steer the spacecraft to a final target state despite the considered perturbations. The control policy is approximated by a deep recurrent neural network, trained by policy-gradient reinforcement learning on a collection of environments featuring mixed sources of uncertainty, namely dynamic uncertainty and control execution errors. The recurrent network is able to build an internal representation of the distribution of environments, thus better adapting the control to the different stochastic scenarios. The results in terms of optimality, constraint handling, and robustness on a fuel-optimal low-thrust transfer between Earth and Mars are compared with those obtained via a traditional reinforcement learning approach based on a feed-forward neural network
Indirect optimization of finite-thrust cooperative rendezvous
No full textA time-constrained finite-thrust rendezvous between two cooperating spacecraft is investigated in detail in this paper. To ensure high numerical accuracy, the optimization is carried out by means of an indirect method that exploits an “a priori” subdivision of the trajectory into burn and coast arcs, whose time lengths become additional
unknown parameters. Issues related to simultaneous presence of two switching control structures, one for each maneuvering spacecraft, are analyzed. A peculiar approach, which relies on the adoption of a different timescale for each spacecraft, is proposed; the introduction of multiple decision vectors allows for an easy management of the
switching control structures for missions involving two or more maneuvering spacecraft. As an example, a coplanarrendezvous is thoroughly analyzed; necessary conditions for optimality are comprehensively derived for both target/chaser and cooperative maneuvers. Numerical results are provided and solutions are discussed
Reinforcement learning for robust trajectory design of interplanetary missions
No full textThis paper investigates the use of reinforcement learning for the robust design of low-thrust interplanetary trajectories in presence of severe uncertainties and disturbances, alternately modeled as Gaussian additive process noise, observation noise, and random errors in the actuation of the thrust control, including the occurrence of a missed thrust event. The stochastic optimal control problem is recast as a time-discrete Markov decision process to comply with the standard formulation of reinforcement learning. An open-source implementation of the state-of-the-art algorithm Proximal Policy Optimization is adopted to carry out the training process of a deep neural network, used to map the spacecraft (observed) states to the optimal control policy. The resulting guidance and control network provides both a robust nominal trajectory and the associated closed-loop guidance law. Numerical results are presented for a typical Earth–Mars mission. To validate the proposed approach, the solution found in a (deterministic) unperturbed scenario is first compared with the optimal one provided by an indirect technique. The robustness and optimality of the obtained closed-loop guidance laws is then assessed by means of Monte Carlo campaigns performed in the considered uncertain scenarios
Comparative Analysis of Reinforcement Learning Algorithms for Robust Interplanetary Trajectory Design
No full textThis paper focuses on the application of reinforcement learning to the robust design of low-thrust interplanetary trajectories in presence of severe dynamical uncertainties modeled as Gaussian additive process noise. A closed-loop control policy is used to steer the spacecraft to a final target state despite the perturbations. The control policy is approximated by a deep neural network, trained by reinforcement learning to output the optimal control thrust given as input the current spacecraft state. The effectiveness of three different model-free reinforcement learning algorithms is assessed and compared on a three-dimensional low-thrust transfer between Earth and Mars elected as study case
Covariance control for stochastic low-thrust trajectory optimization
No full textThis paper outlines a novel approach to the design of optimal space trajectories under significant uncertainty. Finite-horizon covariance control, i.e., the steering of a system from an initial probability distribution to a desired one at a prescribed time, is employed to plan an optimal nominal path along with a robust feedback controller that compensates for exogenous in-flight disturbances. The major contribution of the present paper is a mindful convexification strategy to recast the nonlinear covariance control problem as a deterministic convex optimization problem. The convexification is based on a convenient change of variables that allows to relax the covariance matrix discrete-time propagation into a set of semidefinite cone constraints. While featuring a larger feasible space, the relaxed problem shares the same optimal solution as the original one, as proven by numerical experiments, hence demonstrating that the proposed relaxation is lossless. Monte Carlo campaigns are carried out to validate the in-flight performance of the attained control policies
Robust Waypoint Guidance of a Hexacopter on Mars using Meta-Reinforcement Learning
No full textThis paper presents a meta-reinforcement learning approach to the robust and autonomous waypoint guidance of a six-rotor unmanned aerial vehicle in Mars' atmosphere. The meta-learning is implemented by using a recurrent neural network as a control policy to map data about the hexacopter state provided by onboard sensors to the six rotor angular speeds. The network is trained by proximal policy optimization, a state-of-the-art policy gradient reinforcement learning algorithm. During the training, the network is also provided with information about the previous control output and reward, to improve the policy adaptability to different environment instances. Several mission scenarios, involving uncertainties on Mars' atmosphere's properties, the presence of random wind gusts, and Gaussian noise on the collected sensor data, are investigated to assess the robustness of the proposed approach in realistic operative conditions. The flexibility and performance of meta-reinforcement learning are also compared against standard reinforcement learning with a fully-connected neural network, to better highlight the potential of the proposed methodology in real-world autonomous guidance applications
Convex Optimization of Ascent and Powered Descent of a Reusable Launch Vehicle
No full textThis paper presents a convex programming approach for the optimization of the full ascent trajectory of reusable launch vehicles, from lift-off to orbit payload injection, together with the soft landing of the first stage. A combination of lossless and successive convexification methods is employed to handle the nonlinear dynamics and constraints. Two strategies for the recovery of the first stage, that is, downrange landing and return-to-launch site, are discussed. Preliminary results are presented to show the effectiveness and performance of the proposed approach for a study case involving a two-stage launch vehicle
Adaptive attitude control of launch vehicles in atmospheric flight
No full textThe inclusion of an adaptive augmenting control (AAC) component in the flight control system (FCS) of launch vehicles can be highly effective for enhancing control stability and robustness with respect to parametric uncertainties and dealing with off-nominal conditions, so as to extend the envelope of failures and flight anomalies that can be managed by the vehicle control system. In this paper the adoption of an adaptive notch filter in a control architecture consisting of proportional-derivative (PD) elements, bending filters and AAC is proposed and discussed. The main goal of the study is to investigate the feasibility of implementation and the possible benefits of filter adaptation, such as overcoming critical limitations that degrades the AAC effectiveness for large uncertainties on elastic mode characteristics. To this end, the frequency of first bending mode is estimated during the flight in order to adapt the design parameters of the notch filter. Adaptive control performance is evaluated by simulation of vehicle motion in the atmospheric flight phase in selected stressing cases. Results of Monte Carlo simulations are also discussed for a broader assessment of the effects of adaptive filter on the robustness of integrated FCS
Spacecraft dynamics under the action of Y-dot magnetic control law
Full text linkThe paper investigates the dynamic behavior of a spacecraft when a single magnetic torque-rod is used for achieving a pure spin condition by means of the so-called Y-dot control law. Global asymptotic convergence to a pure spin condition is proven on analytical grounds when the dipole moment is proportional to the rate of variation of the component of the magnetic field along the desired spin axis. Convergence of the spin axis towards the orbit normal is then explained by estimating the average magnetic control torque over one orbit. The validity of the analytical results, based on some simplifying assumptions and approximations, is finally investigated by means of numerical simulation for a fully non-linear attitude dynamic model, featuring a tilted dipole model for Earth׳s magnetic field. The analysis aims to support, in the framework of a sound mathematical basis, the development of effective control laws in realistic mission scenarios. Results are presented and discussed for relevant test cases
- …
