Melkior Ornik
University of Illinois at Urbana-Champaign
ESI18 Melkior Ornik Quad Chart
Intelligent systems in space naturally operate in partly unknown environments, presenting a major challenge to the presently available methods for mission and motion planning. While currently existing strategies do allow systems to adapt their plan in real time in line with the results observed during the mission, capabilities of such strategies are limited: they do not consider heterogeneous information available from previous or parallel missions, nor approach system safety constraints and unexpected situations in a formal way. Instead, the onus is on a human supervisor to deal with any issues in an ad hoc manner. The proposed effort introduces a novel framework of safety-constrained, efficient, and resilient operation for intelligent systems in space. It will develop methods allowing for formal safety specifications to be naturally included in planning algorithms and use them ensure system safety while the system is learning about its environment. Additionally, it will introduce a formal mechanism for describing and exploiting prior and side information available to the system in order to expedite learning. Finally, it will address unexpected situations occurring during the mission by taking a proactive, instead of reactive, approach. Namely, the proposed effort will provide a formal framework for a priori design of intelligent systems that guarantees continued system operation under a wide class of unexpected adverse scenarios. While the effort’s case study will describe a scenario of a lunar rover, the theory and algorithms developed in the proposed effort will be able to serve in support of a range of future missions.