Qin Yang

My research focuses on Distributed Artificial Intelligence (DAI), Swarm Intelligence, Multi-Agent Systems (MAS), and Multi-Robot Systems (MRS). In the dissertation research, I first propose a principled MAS cooperation framework, Self-Adaptive Swarm System (SASS) to bridge the fourth level automation gap between perception, communication, planning, execution, decision-making, and learning. Following the general framework, we propose Robot Needs Hierarchy to model the intelligent agent's motivation and offer a priority queue in a distributed Negotiation-Agreement Mechanism avoiding plan conflicts effectively. Then, we provide several Atomic Operations to decompose the complex tasks into a series of simple sub-tasks. Furthermore, we introduce a network model called Game-theoretic Utility Tree (GUT) mimicking the intelligent agent decision-making process for MAS cooperation in uncertain environments, especially in adversarial scenarios. Besides, to analyze the reliability and stability of relationships between agents or groups in MAS cooperation, we define the needs-based agent trust model -- Relative Needs Entropy (RNE) -- to improve the decision-making and learning performance in MAS cooperation. Moreover, we provide the GUT Bayesian Adaptive Learning to organize agents' behaviors and optimize their strategies efficiently and balance the utility between individual and group reasonably to fulfill agents' highest level needs learning helping SASS self-evolution. To evaluate our models, we propose the Explore Domain and a new application in urban search and rescue (USAR) missions.