Building Space Infrastructure as a Service: An Automated Planning and Scheduling System for a Heterogeneous Spacecraft Constellation

Loft Orbital deploys space infrastructure as a service. With the advent of smallsat technology and the fast growth of the space-as-a-service market, this infrastructure aims at supporting a broad range of applications for a variety of customers, involving a multiplicity of spacecraft platforms and payload types. In order to operate these heterogeneous space systems, while providing high controllability and monitoring capacity to its customers, Loft has developed a fully-automated, web-based mission control system (MCS), called Cockpit. Designed to grant access and control of on-orbit assets to operators as well as customers, Cockpit can be leveraged by endusers to send requests to task their payloads. Predicated on a set of temporal and/ or geospatial constraints and service-level agreement (SLA) based policies, these requests can be formulated either programmatically, leveraging Cockpit's application programming interface (API), or interactively, using its graphical user interface (GUI). This entity is thus responsible for coordinating and automating end-to-end mission planning and scheduling, from end-user requests ingestion and processing, schedule management and optimization, to tasking and execution monitoring. The diversity of space systems being tasked, the complexity of predicting and simulating their state over long-time horizons, the flexibility and the deconfliction capability required by an open requesting engine, and the curse of dimensionality of the scheduling problem pose unique mission planning and scheduling challenges. In order to respond to these challenges, Cockpit's mission management services implement abstractions based on simple and generic requesting and scheduling concepts. This makes for a mission-agnostic simulation, processing and tasking core, combined with mission-specific configurations and resolvers, fostering software scalability and maintainability. In addition, and without compromising with observability of the space systems being tasked, these abstractions promote computational tractability of the scheduling problem-two prerequisites for performing complex simulation, resource allocation and planning optimization. This paper presents some of the key aspects and tradeoffs of the aforementioned technical challenges and how Cockpit's mission planning system is built to achieve automated planning and scheduling of Loft's heterogeneous constellation.