IntelliChair: An Approach for Activity Detection and Prediction via Posture Analysis

Abstract

This paper proposes a robust, low-cost, sensor based system that is capable of recognising sitting postures and placing them in correspondence with sitting activities. This system is also capable of predicting subsequent activities for individual users. Force Sensing Resistors are mounted on the seat and back of a chair to gather the hap tic (i.e., Touch-based) posture information. Subsequently, posture information is fed into two classifiers, one for back posture and the other one for leg posture. A hidden Markov model approach is used to establish the activity model from sitting posture sequences. Furthermore, by implementing a context awareness prediction algorithm (e.g. Active-Lezi), the system discovers patterns and predicts subsequent activities. The system will lead to many potential applications such as the analysis of sitting or lying subjects, motion tracking for rehabilitation, interaction assistance, and the detection of anomalous activities.