Resilient Multi-Agent Collaborative Spacecraft Inspection

Distributed spacecraft systems (DSS) involving SmallSats in low Earth orbit are gaining significant interest both for Earth observation and on-orbit servicing purposes. However, the miniaturized low-cost components of SmallSats are susceptible to faults, making DSS prone to failures that are detrimental to its overall system performance. In this work, we address the problem of providing resiliency to potential failures for a fleet of spacecraft that are performing on-orbit inspection. The proposed methodology guarantees graceful degradation of the inspection performance even against the worst-case failures, through selection and assignment of formation orbits that are resilient to it. We define quantitative metric to measure collaborative inspection performance, taking into consideration both the information gain and control cost, and formulate worst-case failure of ℓ-spacecraft that maximally undermine it. The main algorithm searches through the space of formation orbits, sampling through a set of orbits with evaluation of its inspection performance in the presence of worst-case failures. The algorithms are designed to be computationally efficient and have linear scaling to the number of spacecraft and orbits, making it applicable to real-time planning for spacecraft swarms. The effectiveness of the proposed approach is validated through simulation experiments on a design reference mission involving five CubeSats inspecting a target spacecraft in low Earth orbit.