A Hybrid Reasoning Approach for Activity Recognition Based on Answer Set Programming and Dempster–Shafer Theory

This chapter discusses a promising approach for multisensor-based activity recognition in smart homes. The research originated in the domain of active and assisted living, particularly in the field of supporting people in mastering their daily life activities. The chapter proposes (a) a reasoning method based on answer set programming that uses different types of features for selecting the optimal sensor set, and (b) a fusion approach to combine the beliefs of the selected sensors using an advanced evidence combination rule of Dempster–Shafer theory. In order to check the overall performance, this approach was tested with the HBMS dataset on an embedded platform. The results demonstrated a highly promising accuracy compared to other approaches.