Hidden Markov Models for Activity Recognition in Ambient Intelligence Environments

Context-aware computing offers several advantages for human computer interaction by augmenting ambient intelligence environments with computational artifacts that can be responsive to the needs of users. One of the main challenges in context-aware computing is context recognition. While some contextual variables, such as location, can be easily recognized, others, such as activity are more complex to estimate. This paper describes an approach to estimate activities in a working environment. The approach is based on information gathered from a workplace study, in which 196 hours of detailed observation of hospital workers were recorded. This data is used to train a Hidden Markov Model to estimate user activity. The results indicate that the user activity can be correctly estimated 92.6% of the time. We compare our results with the use of neuronal networks and human observers familiar with those work practice. We discuss how these results can be used for context-aware applications.