Toward an affective intelligent machine system: theory and application

智能机器系统是指能够展示人类的特征的机器，如学习、推理、决策、感知和适应。现代智能机器系统排除了情感，这从根本上损害了系统内的智能程度，因为情感影响着我们的信息处理、决策和记忆能力。这项研究计划旨在更好地理解情绪对我们的决策过程有什么影响以及如何影响。研究将在两个典型的应用领域(即驾驶员-车辆系统和老年人-机器人系统)进行，以展示拟议研究的普遍性；之所以选择它们，是因为它们对我们的社会产生了广泛的影响。第一个问题是，当他或她与之交互的机器有情感并表现出它的情感时，会对人类的行为产生什么重大影响。第二个问题是，机器和人类如何共同控制人机系统，以进一步提高其性能。我们相信，拟议的研究将产生知识来回答这些问题。具体地说，我们将首先开发一种方法，将机器的情感表达构建为与人类的相同，并研究这种情感表达如何影响人类的情感和认知。其次，我们开发了几个模型来表示人与机器交互时的情绪和表现之间的关系。最后，我们为机器开发方法，以智能地帮助人类控制人机系统；辅助包括更多地了解机器中发生的事情，提供建议信息，并在需要时对人类采取行动。从智力上讲，这项研究将进一步推进我们在构建智能机器系统方面的知识。这项研究将对公共安全和健康产生巨大的社会效益。

Intelligent machine systems refer to machines that can exhibit the traits found in humans, such as learning, reasoning, decision-making, sensing, and adaptation. The contemporary intelligent machine system has excluded emotion, which thus fundamentally compromises the degree of intelligence within the system because the emotion affects our capability of information processing, decision making, and memory. This research program aims to develop a better understanding of what and how the emotion affects our decision making process. The research will be conducted in two typical application domains (i.e., driver-vehicle and elderly-robot systems) in order to demonstrate the generality of the proposed research; they are chosen due to their broad impact to our society.    There are two general questions to be answered through this research program. The first question is what would be some significant effect on the human behavior when the machine he or she interacts with has emotion and shows its emotion. The second question is how the machine and human co-control a human-machine system to further improve its performance.    The proposed research will generate knowledge to answer these questions. Specifically, we will first develop a methodology to construct the emotional expression of a machine to be the same as that of a human and examine how this emotional expression affects the human emotion and cognition. Second, we develop several models that can represent the relationship between the emotion and performance of the human when he or she interacts with the machine. Last, we develop methodologies for the machine to intelligently assist the human in controlling the human-machine system; assistances include more understanding of what happens in the machine, offering advisory message, and taking actions over the human when needed.    Intellectually, this research will further advance our knowledge on building intelligent machine systems. This research will generate great social benefits to public safety and health.