Towards understanding expression for tele-operation

Human expressions contain important information during communication. Expressions are often used to quickly understand the basic underlying intent of a message being conveyed. This paper presents an approach that leverages human expression for remote tele-operation tasks and to augment shared multiparticipant environments with meaningful concepts. Taking advantage of this information helps to minimise the human time required to convey intent. Expressions are observed through hand gestures and facial expressions, basic primitives are identified using a fuzzy-hidden Markov model approach and sets of these primitives are used to infer intent using a domain specific conceptual-graph based knowledge system. Although dynamic hand gestures and basic facial expressions are used as sources of human expression, the flexibility exists to incorporate additional and alternate sources of human expression. The proposed approach to identify meaningful concepts from human expression can be a valuable tool in a multiparticipant collaborative environment. Multiparticipant multimedia collaboration benefits from this computer-assisted understanding approach, as culture-specific expressions can be automatically clarified to reduce ambiguity and misunderstanding.