A Framework for UAV Swarm Situation Awareness

This paper proposes a framework for situation awareness in a multi-agent environment. In our work, the framework is composed of 3 key components: situation elements identification, situation comprehension, and situation prediction. Taking the example of a UAV swarm engaged in search and rescue operations. Firstly, we study the UAV swarms from a holistic perspective and define situation elements as the formation and motion characteristics of swarm. The Multi-Layer Perceptron (MLP) neural network is designed to identificate the swarm collective behavior formation. Then, this paper proposes a fuzzy inference system to comprehend the current situation of the swarm, deducing the current behavioral intentions. Finally, the Long-Short Term Memory (LSTM) neural network is employed to predict the future situation of swarm. And the advantage assessment function is given to quantitatively analyze the swarm’s advantage in the search and rescue operation.