A Framework for Anomaly Detection in Activities of Daily Living using an Assistive Robot

Abstract

This paper presents an overview of an ongoing research to incorporate an assistive robotic platform towards improved detection of anomalies in daily living activities of older adults. This involves learning human daily behavioural routine and detecting deviation from the known routine which can constitute an abnormality. Current approaches suffer from high rate of false alarms, therefore, lead to dissatisfaction by clients and carers. This may be connected to behavioural changes of human activities due to seasonal or other physical factors. To address this, a framework for anomaly detection is proposed which incorporates an assistive robotic platform as an intermediary. Instances classified as anomalous will first be confirmed from the monitored individual through the intermediary. The proposed framework has the potential of mitigating the false alarm rate generated by current approaches.