Narratives for Multi-Robot Event Recognition – Flinders/ENIB, Funding: Naval Group Pacific.
Summary: The aim of this project is to develop an event recognition system that utilises logic-based reasoning techniques and machine learning methods to interpret joint actions executed by a team of autonomous maritime vehicles. This system aims to analyse the behaviour and intentions of individual robots within a collective setting. To achieve this, the project involves several key steps. First, an ontology will be constructed to represent high-level joint actions and their relationships, serving as the knowledge base for logical reasoning. Machine learning algorithms will be incorporated to train models that can recognise and classify joint actions based on sensor data and perceptual information. Additionally, the system will incorporate temporal reasoning mechanisms to understand the sequence and timing of joint actions. This capability will allow the system to identify causal relationships and dependencies between actions, enabling the generation of concise summaries of the vehicle’s missions. To evaluate the system’s effectiveness, extensive experiments will be conducted using a simulated multi-robot environment. Evaluation metrics will include accuracy in recognising joint actions, summarisation quality, and computational efficiency.