Asteroid deflection campaign design integrating epistemic uncertainties

Asteroid deflection entails multiple sources of epistemic uncertainties and stochastic uncertainties. Epistemic uncertainties can be reduced by replenishing incomplete information with a better means of observation, whereas stochastic uncertainties are inherent and irreducible owing to their randomness. Our understanding of asteroid deflection is largely limited by lack of in-situ data or full-scale experiments. Reducing these epistemic uncertainties in (1) the physical properties of asteroids or (2) the consequences of human interference will drastically benefit our mitigation efforts against their hazards. Much of previous literature concentrated on single-stage missions or simple asteroid models with fixed physical attributes, making it inadequate to handle epistemic or stochastic uncertainties. To fill this gap, this paper tries to incorporate different kinds of uncertainties within a campaign design framework. In a multi-stage campaign, a precursor is deployed first to reduce epistemic uncertainties. Because stochastic uncertainties cannot be reduced by nature, the campaign should be optimized to maximize its robustness against the worst random situations. Finally, whether or not a two-stage campaign is better than a single-stage mission can be visualized as a decision map for decision-makers. The paper analyzes a hypothetical scenario of deflecting 99942 Apophis and discusses future work.