用户状态隐式感知与智能干预方法研究

The essence of Human-Computer Interaction is the two-way communication between humans and computers. However, it is challenging to keep the information flow balanced between such communications with traditional graphical user interfaces. The increasing computational power of computers make it possible to generate and output vast amounts of data while they can still only infer users' intentions via analyzing users’ explicit and discrete interactive behaviors. Systems still have a limited ability to detect the full spectrum of information that is naturally and effortlessly generated by users during computer usage. The lack of such information makes it difficult to dynamically adapt to users and achieve user-centric computing. The goal of this research is to enhance the communication bandwidth between humans and machines via implicit user state detection and intelligent interface-level interventions. We will explore and investigate the relationship between human's existing interactive behaviors and physiological signals and their physical functions, as well as cognitive and affective states. We will make technical breakthroughs in domains such as signal processing, multi-modal information fusion and implicit perception of user intentions. Moreover, we will research and propose design principles and design space of interventions and adaptations based on such implicit input from users. Finally, we will build prototyping systems and conduct case studies. This research has great significance to promote the development of human-machine symbiosis and to achieve the ultimate goal of natural interaction between humans and machines.

人机交互的本质是人与计算机的双向交流。然而，在图形用户界面等传统交互环境中，用户和计算机的交流很难做到信息对等。日益提升的处理能力使计算机可以向用户快速地输出大量结果和状态信息，而计算机仅能够通过用户主动离散的交互行为推断其交互意图，并不能获取、分析和利用用户自然流露的状态信息。这些信息的缺失使得系统很难做到动态适应用户，真正实现以人为中心的计算。本项目旨在通过隐式的用户状态感知和智能的界面干预拓宽人机交互带宽，提高人机系统的工作效率。本项目将探索和揭示隐含于用户已经建立的交互行为和生理信号之中的特征与身体机能、认知及情感状态间的相互作用机理，突破信号处理、多通道融合、隐式用户意图理解等关键技术，研究并提出基于非精确用户状态输入的界面干预设计原则和设计空间，并面向关键领域构建典型应用。该研究对于促进人机交互方式由主从关系向协同共生关系发展、实现真正的自然人机交互具有重要意义。

人机交互的本质是人与计算机的双向交流。然而，在图形用户界面等传统交互环境中，用 户和计算机的交流很难做到信息对等。日益提升的处理能力使计算机可以向用户快速地输出大 量结果和状态信息，而计算机仅能够通过用户主动离散的交互行为推断其交互意图，并不能获取、分析和利用用户自然流露的状态信息。这些信息的缺失使得系统很难做到动态适应用户， 真正实现以人为中心的计算。本项目通过研究隐式的用户状态感知和智能的界面干预拓宽人机交互带宽，提高人机系统的工作效率。本项目探索和揭示隐含于用户已经建立的交互行为和生理信号之中的特征与身体机能、认知及情感状态间的相互作用机理，突破信号处理、多通道 融合、隐式用户意图理解等关键技术，研究并提出基于非精确用户状态输入的界面干预设计原 则和设计空间，并面向医疗领域构建典型应用。在该项目支持下，项目组顺利完成既定的各项研究目标。共发表高质量学术论文5篇，其中CCF-A类论文4篇（1篇获得人机交互领域顶级会议ACM CHI 2019最佳论文提名奖），CCF-B类论文1篇；申请专利3项；培养研究生4名（其中博生生3名，硕士生1名）。