Human Agent Path-Planning for Spacecraft Motion with Deterministic Chaos, Small Random Perturbations and Random Parameters

摘要

Small body space environments may be subjected to coupled interactions of chaotic dynamics, small random perturbations, and unknown system parameters (CD-SRP-UP). These interactions pose significant challenges to autonomous spacecraft path-planning schemes. Currently, path-planning strategies mostly rely on baselined solutions that are derived from known orbit dynamical models by human experts. However, a baselined path-planning approach may fail to deliver, in advance, feasible solutions when the dynamical model is poorly known, such as in the case of small body systems. Deterministic chaos and bifurcations within orbit dynamics may also hinder extensions of existing baselined solutions to different mission environments that are governed by identical equations of motion but different system parameters. Rather than relying on the automatization of the search process for natural, flow-compliant trajectories for each new system dynamics, autonomous spacecraft guidance within CD-SRP-UP dynamics may be implemented via baseline-free strategies. Baseline-free guidance may stem from demonstration based learning via observations of a human agent controlling simulated spacecraft motion. In this work, we focus on collecting and analysing human agent path planning strategies within simulated CD-SRP-UP dynamics. The flight simulator renders orbit dynamics via an Elliptical Restricted Three-Body Problem (ER3BP) model with the addition of perturbations caused by irregular gravity and solar radiation pressure. Small random perturbations are simulated by orbit determination errors, and unknown system parameters are incorporated by the random generation of the system parameters for each simulation run. The resulting spacecraft motion is chaotic and subjected to high degrees of uncertainties; therefore, spacecraft path-planning in such an environment may benefit from human inspired guidance. A previous numerical experiment collected initial evidence for the feasibility of human agent