State-of-the-art weather forecasting systems depend on a variety of data collected by airborne, orbiting, and ground sensors. Regional CubeSat constellations have the potential to improve hurricane forecasting by collecting sensor data over data-starved oceanic regions. Even in regions where strong terrestrial sensor networks exist, constellation sensor data can help reduce forecasting model errors. To this end, the article considers the problem of designing a low-earth orbit CubeSat constellation that meets given resolution requirements over a region of interest. We propose a novel optimization framework that uses the concept of satellite coverage maps to determine the number of satellites and constellation pattern. Numerical simulations are presented for asymmetric constellation design that can provide sensor data over important geographical regions within a specified repeated time window.
Transferring an all-electric spacecraft from a launch injection orbit to the geosynchronous equatorial orbit (GEO) using a low thrust propulsion system presents a significant challenge due to the long transfer time typically spanning several months. To address the challenge of determining such long time-scale orbit-raising maneuvers to GEO, this paper presents a novel technique to compute transfers starting from geostationary transfer orbit (GTO) and super-GTO. The transfer is complex, involving multiple eclipses and revolutions. To tackle this challenge, we introduce a cascaded deep reinforcement learning (DRL) model to guide a low-thrust spacecraft towards the desired orbit by determining an appropriate thrust direction at each state. To ensure mission requirements, a gradient-aided reward function incorporating the orbital elements, guides the DRL agent to obtain the optimal flight time. The obtained results demonstrate that our proposed approach yields optimal or near-optimal time-efficient spacecraft orbit-raising. DRL implementation is important for spacecraft autonomy; in this context, we demonstrate that our DRL-based trajectory planning provides significantly better transfer time as compared to state-of-the-art approaches that allow for automated trajectory computation.
In this paper, we consider a recently developed formulation of the electric orbit-raising problem that utilizes a novel dynamic model and a sequence of optimal control sub-problems to yield fast and robust computations of low-thrust trajectories. This paper proposes two enhancements of the computational framework. First, we use thruster efficiency in order to determine the trajectory segments over which the spacecraft coasts. Second, we propose the use of a neural network to compute the solar array degradation in the Van Allen radiation belts. The neural network is trained on AP-9 data and SPENVIS in order to compute the associated power loss. The proposed methodology is demonstrated by considering transfers from different geosynchronous transfer orbits. Numerical simulations analyzing the effect of thruster efficiency and average power degradation indicate the suitability of starting the maneuver from super-geosynchronous transfer orbits in order to limit fuel expenditure and radiation damage. Furthermore, numerical simulations demonstrate that proposed enhancements are achieved with only marginal increase in computational runtime, thereby still facilitating rapid exploration of all-electric mission scenarios.