Optimization-Based Task Allocation for Earth Observation in Multi-Satellite Systems

The increasing deployment of CubeSats and small satellite constellations for Earth observation has enabled more agile and responsive systems and data services. It has also introduced new challenges related to limited onboard resources, diverse sensor modalities, and the need for dynamic task allocation. Efficient coordination in heterogeneous multi-satellite systems, comprising platforms with differing orbital configurations and imaging capabilities such as optical or radar, requires planning approaches that account for these constraints while minimizing communication overhead and latency. In this paper, we propose a optimization-based task allocation framework that enables heterogeneous, constraint-aware and globally optimal task assignment without relying on inter-satellite communication. Leveraging a hypergraph-based representation to model the compatibility between observation tasks and satellite capabilities, we formulate the allocation problem as a mixed-integer linear program (MILP) that directly incorporates key operational constraints such as energy consumption, data storage, and inter-task timing dependencies. We evaluate the effectiveness of the method using a realistic Earth observation scenario involving the HYPSO-1 and HYPSO-2 hyperspectral CubeSats. Furthermore, we extend the formulation to support secondary objectives, such as maximizing the coverage of an area of interest and minimizing data latency to ground stations, where our method outperforms baseline heuristics.