Search

In this paper, we introduce a novel deep learning based solution to the Powered-Descent Guidance (PDG) problem, grounded in principles of nonlinear Stochastic Optimal Control (SOC) and Feynman-Kac theory. Our algorithm solves the PDG problem by framing it as an $\mathcal{L}^1$ SOC problem for minimum fuel consumption. Additionally, it can handle practically useful control constraints, nonlinear dynamics and enforces state constraints as soft-constraints. This is achieved by building off of recent work on deep Forward-Backward Stochastic Differential Equations (FBSDEs) and differentiable non-convex optimization neural-network layers based on stochastic search. In contrast to previous approaches, our algorithm does not require convexification of the constraints or linearization of the dynamics and is empirically shown to be robust to stochastic disturbances and the initial position of the spacecraft. After training offline, our controller can be activated once the spacecraft is within a pre-specified radius of the landing zone and at a pre-specified altitude i.e., the base of an inverted cone with the tip at the landing zone. We demonstrate empirically that our controller can successfully and safely land all trajectories initialized at the base of this cone while minimizing fuel consumption.

We present a general framework for optimizing the Conditional Value-at-Risk for dynamical systems using stochastic search. The framework is capable of handling the uncertainty from the initial condition, stochastic dynamics, and uncertain parameters in the model. The algorithm is compared against a risk-sensitive distributional reinforcement learning framework and demonstrates outperformance on a pendulum and cartpole with stochastic dynamics. We also showcase the applicability of the framework to robotics as an adaptive risk-sensitive controller by optimizing with respect to the fully nonlinear belief provided by a particle filter on a pendulum, cartpole, and quadcopter in simulation.

Liquid propellant makes up a significant portion of the total weight for large launch vehicles such as Saturn V, Space Shuttle, and the Space Launch System. Careful attention must be given to the influence of fuel slosh motion on the stability of the vehicle in the design of the Flight Control System (FCS). Historically, there have been instances where a poorly designed FCS in addition to a lack of passive damping have caused the slosh mass to drive the launch vehicle unstable. The dynamics behind controlling a quadcopter/hanging mass configuration is analogous to that of controlling the attitude of a rocket with a single propellant tank. The quadcopter/hanging mass configuration offers a reasonably accurate platform for assessing the real-time effectiveness of the SLS Adaptive Augmenting Controller in suppressing slosh instability. Flight test experiments were carried out at the NASA Langley Research Center's Autonomy Incubator. During both simulation and flight test, the hanging mass was intentionally made unstable and the adaptive algorithm successfully suppressed the instability as expected.