Creating Interactive Virtual Humans: Some Assembly Required

摘要

Science fiction has long imagined a future populated with artificial humans- human-looking devices with human-like intelligence. Although Asimov's benevolent robots and the Terminator movies' terrible war machines are still a distant fantasy, researchers across a wide range of disciplines are beginning to work together toward a more modest goal building virtual humans. These software entities look and act like people and can engage in conversation and collaborative tasks, but they live in simulated environments. With the untidy problems of sensing and acting in the physical world thus dispensed, the focus of virtual human research is on capturing the richness and dynamics of human behavior. This broad range of requirements poses a serious problem. Researchers working on particular aspects of virtual humans cannot explore their component in the context of a complete virtual human unless they can understand results across this array of disciplines and assemble the vast range of software tools (for example, speech recognizers, planners, and animation systems) required to construct one. Moreover, these tools were rarely designed to interoperate and, worse, were often designed with different purposes in mind. For example, most computer graphics research has focused on high fidelity offline image rendering that does not support the fine-grained interactive control that a virtual human must have over its body. In the spring of 2002, about 30 international researchers from across disciplines convened at the University of Southern California to begin to bridge this gap in knowledge and tools. Our ultimate goal is a modular architecture and interface standards that will allow researchers in this area to reuse each other's work. This goal can only be achieved through a close multidisciplinary collaboration. Towards this end, the workshop gathered a collection of experts representing the range of required research areas. Here we discuss some of the key issues.