Autonomous Decision-Making With Incomplete Information and Safety Rules Based on Non-Monotonic Reasoning

In this article we propose a decision process integrating Non-Monotonic Reasoning (NMR), embedded in deliberative architecture. The NMR uses Default Logic to implement goal reasoning, managing partially observable or incomplete information, allowing the design of default behaviours completed by handling specific situations, order manage current mission objective as well safety rules. We illustrate our approach through an application underwater robot performing marine biology mission.