I. E. Yairi, T. Yairi, and S. Igi (Japan)
Pattern Recognition, Machine Learning, Sup port Vector Machine, Micro-level Intentions, Multi dimensional Time-series
Real-time intention recognition by sensing a user and its environment is an important function for man machine interaction. However, conventional methods are not applicable to the recognition of micro-level in tentions, because it is difficult to divide a user's micro level behavior into a clear sequence of action fractions and it varies from person to person. Our approach to this problem is to apply a pattern learning function which discovers and utilizes synchronous and tempo ral relations among the multi-dimensional time-series data of the user and environment sensors. In this pa per, we deal with a vehicle driving task as a typical ap plication of the proposed intention recognition method, and examine the realizability of the pattern learning function by SVM. Importance of sensing both a user and the environment for intention recognition in vehi cle driving is discussed as well as our future plan.
Important Links:
Go Back
