This paper investigates the use of reinforcement learning for robust design low-thrust interplanetary trajectories in presence severe uncertainties and disturbances, alternately modeled as Gaussian additive process noise, observation random errors actuation thrust control, including occurrence a missed event. The stochastic optimal control problem is recast time-discrete Markov decision to comply with standard formulation learning. An open-source implementation state-of-the-art algorithm...

This paper investigates the use of deep learning techniques for real-time optimal spacecraft guidance during terminal rendezvous maneuvers, in presence both operational constraints and stochastic effects, such as an inaccurate knowledge initial state random in-flight disturbances. The performance two well-studied methods, behavioral cloning (BC) reinforcement (RL), is investigated on a linear multi-impulsive mission. To this aim, multilayer perceptron network, with custom architecture,...

This paper deals with the optimization of ascent trajectory a multistage launch vehicle, from liftoff to payload injection into target orbit, considering inverse-square gravity acceleration and aerodynamic forces. A combination lossless successive convexification techniques is adopted generate sequence convex problems that rapidly converges original problem solution. An automatic initialization strategy proposed make solution process completely autonomous. In particular, novel three-step...

This paper outlines a novel approach to the design of optimal space trajectories under significant uncertainty. Finite-horizon covariance control, i.e., steering system from an initial probability distribution desired one at prescribed time, is employed plan nominal path along with robust feedback controller that compensates for exogenous in-flight disturbances. A mindful convexification strategy devised recast nonlinear control problem as deterministic convex optimization problem. The based...

This paper presents a convex programming approach to the optimization of multistage launch vehicle ascent trajectory, from liftoff payload injection into target orbit, taking account multiple nonconvex constraints, such as maximum heat flux after fairing jettisoning and splash-down burned-out stages. Lossless successive convexification methods are employed convert problem sequence subproblems. Virtual controls buffer zones included ensure recursive feasibility process, state-of-the-art...

This paper focuses on the use of meta-reinforcement learning for autonomous guidance a spacecraft during terminal phase an impact mission toward binary asteroid system. The control policy is replaced by convolutional-recurrent neural network, which used to map optical observations collected onboard camera thrust and thrusting times. network trained proximal optimization algorithm, family reinforcement methods. final NASA's Double Asteroid Redirection Test (DART) as test case. objective...

This paper presents the main characteristics of evolutionary optimization code named EOS, Evolutionary Optimization at Sapienza, and its successful application to challenging, real-world space trajectory problems. EOS is a global algorithm for constrained unconstrained problems real-valued variables. It implements number improvements well-known Differential Evolution (DE) algorithm, namely, self-adaptation control parameters, an epidemic mechanism, clustering technique, ε-constrained method...

This paper presents a methodology for the concurrent first-stage preliminary design and ascent trajectory optimization, with application to Vega-derived light launch vehicle. The reuse as first stage of an existing upper-stage (Zefiro 40) requires redesign propellant grain geometry, in order account mutated operating conditions. An optimization code based on parallel running several differential evolution algorithms is used find optimal law internal pressure during operation, together thrust...

This paper investigates the use of evolutionary algorithms for optimization time-constrained impulsive multirendezvous missions. The aim is to find minimum- <math xmlns="http://www.w3.org/1998/Math/MathML" id="M1"> <mi>Δ</mi> <mi>V</mi> </math> trajectory that allows a chaser spacecraft perform, in prescribed mission time, complete tour set targets, such as space debris or artificial satellites, which move on same orbital plane at slightly different altitudes. For this purpose, two-level...

This paper investigates the use of neural networks as surrogate models for fast and accurate performance prediction hybrid rocket engines (HREs). To this aim, a deep feedforward network is trained to estimate thrust mass flow rate laws liquid oxygen/paraffin-wax HRE (as well other motor characteristics, including burning time dry propellant masses), starting from limited number design parameters, which define fuel grain, nozzle geometry, initial oxidizer rate. A medium-fidelity ballistic...

Active debris removal missions require an accurate planning for maximizing mission payout, by reaching the maximum number of potential orbiting targets in a given region space. Such problem is known to be computationally demanding and present paper provides technique preliminary based on novel evolutionary optimization algorithm, which identifies best sequence captured and/or deorbited. An original archipelago structure adopted improving algorithm capabilities explore search Several...

This paper investigates the use of reinforcement learning for fuel-optimal guidance a spacecraft during time-free low-thrust transfer between two libration point orbits in cislunar environment. To this aim, deep neural network is trained via proximal policy optimization to map any state optimal control action. A general-purpose reward used guide toward law, regardless specific pair considered and without ad hoc shaping technique. Eventually, learned policies are compared with solutions...

The objective of this work is the preliminary design a low-<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mi>Δ</mml:mi><mml:mi>V</mml:mi></mml:math> transfer from an initial elliptical orbit around Jupiter into final circular moon Europa. This type trajectory represents excellent opportunity for low-cost mission to Europa, accomplished through small orbiter, as in proposed Europa Tomography Probe mission, European contribution NASA’s Multiple-Flyby Mission (or...

A time-constrained finite-thrust rendezvous between two cooperating spacecraft is investigated in detail this paper. To ensure high numerical accuracy, the optimization carried out by means of an indirect method that exploits “a priori” subdivision trajectory into burn and coast arcs, whose time lengths become additional unknown parameters. Issues related to simultaneous presence switching control structures, one for each maneuvering spacecraft, are analyzed. peculiar approach, which relies...

Ground transportation in urban areas faces increasingly severe congestion; however, roads cannot be expanded unlimitedly. Under the air mobility (UAM) concept, electric vertical takeoff and landing (eVTOL) aircraft can use third dimension of space for various needs. However, energy required to complete an UAM mission by eVTOL vehicle must less than available stored battery pack. Furthermore, vehicles fly along feasible paths, such as avoiding collisions with any obstacles arrive at...

This paper presents a convex programming approach for the optimization of full ascent trajectory reusable launch vehicles, from lift-off to orbit payload injection, together with soft landing first stage. A combination lossless and successive convexification methods is employed handle nonlinear dynamics constraints. Two strategies recovery stage, that is, downrange return-to-launch site, are discussed. Preliminary results presented show effectiveness performance proposed study case involving...

This paper presents a convex approach to the design of optimal space trajectories while explicitly accounting for uncertainty. A covariance control problem aimed at driving stochastic system from an initial probability distribution desired one final time is formulated retrieve nominal trajectory and additive state feedback controller that can compensate exogenous in-flight disturbances. The regulates thrust magnitude and, consequently, propellant mass consumption. As result, random variable...