Task Activity Recognition and Workspace Extraction for Nursing Care Assistance in Intelligent Space

With the rapid increase in the elderly population, there is a shortage of caregivers in Japan. Consequently, the workload per caregiver has increased, which is now a major social issue. In response to this, information and communication technology (ICT) systems have been introduced to assist caregivers. However, types of tasks that can be supported are limited. In addition, caregivers need to form a plan of nursing tasks according to various states of care receivers. This requires caregivers to be highly skilled, which in turn increases their cognitive burden. To resolve this, in this study, we aim to realize an assistive system to reduce the cognitive burden of caregivers and improve the efficiency of their performance. We mainly apply the iSpace (intelligent Space) concept to realize the provision of assistance based on the observations of the caregivers and care receivers. Firstly, task activity recognition is required. This paper focuses on task activity recognition. We assume that task activities of caregivers are well related with locations in the nursing environments. Thus, we propose a method to determine a caregiver's task activity based on his or her position and orientation in the potential workspaces defined in advance. In addition, to enhance this approach to wider situations, the potential workspaces should be extracted automatically. For this, we also propose a method to extract potential workspaces of caregivers based on hierarchical clustering of movement trajectories of caregivers. Through experiments, the caregiver's task activity was shown to be accurately recognized and the potential workspace of the caregiver was also successfully extracted.