In-situ Resource Utilisation (ISRU) is considered necessary to enable Off-Earth settlement. It is also regularly compared with terrestrial mining operations. Optimisation of the resource extraction sequence and cut-off grades for product and waste streams are critical for both terrestrial mining operations and ISRU. However, the traditional methods used in the terrestrial mining industry are not directly compatible with ISRU. This paper outlines the differences between terrestrial mining and ISRU and develops a new method for ISRU planning based on Reinforcement Learning (RL). An RL agent is trained and evaluated for extraction sequencing, sometimes showing the ability to outperform a human expert. The generalised RL agent can also be used to run multiple scenarios to determine optimal cut-off grades and conduct risk analysis on varying geological and equipment reliability inputs. Future ISRU projects and assessments will benefit from this method by reducing the human effort required to achieve production optimality.

就地资源利用 (ISRU) 被认为是实现离地定居所必需的。它还定期与陆地采矿作业进行比较。优化产品和废物流的资源提取顺序和边界品位对于陆地采矿作业和 ISRU 都至关重要。然而，陆地采矿业使用的传统方法与 ISRU 不直接兼容。本文概述了陆地采矿和 ISRU 之间的区别，并开发了一种基于强化学习 (RL) 的 ISRU 规划新方法。RL 智能体经过训练和评估以进行提取排序，有时表现出优于人类专家的能力。广义 RL 代理还可用于运行多个场景以确定最佳边界品位，并对不同的地质和设备可靠性输入进行风险分析。未来的 ISRU 项目和评估将受益于这种方法，因为它减少了实现生产优化所需的人力。