Adaptive high degree-of-freedom interaction techniques

点击翻译按钮获取中文摘要

While it is true that computing power and information access has increased by orders of magnitude over the last few decades, it is also true that raw human abilities such as our perceptual and motor systems- astonishing as they are- remain effectively similar from one generation to the next.Yet, the common keyboard/ mouse/ monitor trio is hardly playing to the full potential of human motor and perceptual abilities.The significance of this is that for tasks where today's massive amounts of data and computation are involved,the critical bottleneck in many scientific/creative workflows is now the interface itself-- the bandwidth between the human and machine.As a scientist with extensive professional music experience,I am fascinated by instruments that allow people to perform at an extremely sophisticated level:instruments that require and reward complex cognitive& motor skills and that can be fluidly used to realize sophisticated ideas.I am interested in discovering building blocks of these powerful interaction techniques,and qualities that allow some instruments to be playable in this way.As a computer science researcher I want to create instruments that provide such fluid interaction within the digital realm: tools that let humans naturally drive the immense processing power of today's digital engines to reach their goals and play with their visions more directly.These observations lead to a key question driving my long-term research programme:How do we design interactions in high degree-of-freedom(DOF) audio/ visual/ haptic spaces that allow a user to view and interact effectively with complex data? I explore two different testbeds for data interaction:creating multimedia content and visualizing scientific (aerospace&seismic) data. In both cases,there are shared principles,ideas and techniques for effective and fluid high-DOF interaction,and discovering and identifying these is at the heart of my research.Inasmuch as these are my goals, machine learning algorithms will be a key means of achieving this.Using probabilistic adaptive methods to give the user a wieldy interface over the large complex data sets will be a powerful and practical element of my approach.