Planetary Moon Tours remain among the most complex problems in mission analysis, as they constitute high-dimensional mixed-integer optimisation tasks with virtually unbounded design spaces. Recent studies have proposed automated methods for constructing such sequences, particularly in the context of the Endgame problem, where the objective is restricted to achieving orbit insertion around a target moon. The large trajectory databases pre-computation is followed by dynamic programming exploitation to generate ΔV–Time-of-Flight (ToF) Pareto fronts. This work, conducted within the research activities of the ASTRA lab, extends these approaches by explicitly integrating mission constraints and scientific objectives into the optimisation process, with the goal of automating the design of planetary Moon Tours. Quantitative surface-coverage requirements, maximum flyby altitudes, and limits on cumulative absorbed Total Ionizing Dose (TID) are included as illustrative mission constraints, while the framework itself is general and can accommodate a broad range of design- and operations-related requirements. In addition, the design problem is formulated within a graph framework, enabling the computation of feasible and optimised Tours through the Dijkstra path-planning algorithm.
Trajectory optimisation is therefore combined with scientific requirements and environmental survivability considerations, ensuring that the resulting sequences are feasible in an actual mission context. The proposed pipeline thus integrates feasibility, science-driven performance metrics, and environmental constraints within a unified structure.
The pipeline begins with the automated construction and optimisation of single-moon phases, performed within a Keplerian patched-conics 0-Sphere-of-Influence (SOI) model. This first routine identifies the most favourable transfer sequences, according to a selected optimality criterion, around each moon. The design space is discretised using resonant transfers, non-resonant transfers, V-Infinity Leveraging Transfers (VILTs), Cranking Over the Top (COT) sequences, and classical flybys. All transfers are mapped onto each moon’s v-Infinity Globe, translating the primary-centric velocity vector into the local v-infinity, pump, and crank angle relevant for gravity-assist manoeuvres.
A graph-based representation is adopted to model the entire single-moon transfer design space. Nodes correspond to discrete spacecraft states, defined by v-infinity, pump and crank angle at every moon encounter, while edges represent transfers originating from each node. Each edge is assigned a scalar cost derived from a weighted average of ΔV, ToF, and moon surface coverage during the flybys. Surface coverage is obtained through spherical partition in zones, assigning coverage increments to hyperbolic trajectories, including bounded-altitude segments. By using exact graph-search algorithms such as Dijkstra’s, a systematic exploration of the design space with high computational efficiency is gained, allowing the identification of the optimal transfer sequences with respect to the selected edge weighting function.
Once optimal single-moon phases are obtained, they are patched across moons through inter-moon Lambert arcs, introducing the time-dependence of the problem. Given a departure window, candidate departure dates are obtained by discretising it, and Lambert arcs are computed between the moons in the primary-centred planar frame, with the ToF bounded above by the arrival moon’s orbital period. A second graph structure is subsequently built: in this phase, nodes are extended to include the encounter longitude to preserve the trajectory’s time dependence, while edges represent both Lambert arcs and single-moon solutions treated as transfers. Mission related constraints can be incorporated in the algorithm. As an example, TID constraints have been considered for the Jovian system. These constraints shape the set of admissible solutions, introducing science-driven filtering into early-phase design, enhancing the relevance of the trajectories for realistic mission scenarios.
The pipeline is assessed through case studies in the Jovian and Saturnian systems and through variations in departure windows and cost-function weights, with results also compared against reference trajectories from past and planned missions such as Europa Clipper, thereby demonstrating its performance across diverse mission settings.  The single-moon optimiser autonomously generates high-coverage, low-ΔV flyby sequences for each moon, while the inter-moon patching stage yields multi-moon Tours that exhibit competitive performance compared to sequences known in the literature.
In summary, this work introduces a unified framework for automated scientific Moon Tour design, combining optimised single-moon phases with their dynamical integration through inter-moon transfers. The methodology accommodates diverse mission and science constraints and supports the inclusion of low-energy transfers. The paper illustrates the effectiveness of the proposed structure through a detailed application to two representative scenarios around Saturn and Jupiter, highlighting the capabilities and performance of the full pipeline.