Study of Fall Detection Using Intelligent Cane Based on Sensor Fusion

Abstract

A three-wheeled omni-directional cane robot is designed for aiding the elderly walking. A new human fall detection method is proposed based on fusing sensory information from a vision system and a laser ranger finder (LRF). This method plays an important role in the fall-prevention for the cane robot. The human fall model is represented in a 2D space, where the distance between the head and the average leg position is a significant feature to detect the fall. The possibility distribution of this distance is estimated by using Dubois possibility theory. Fall detection is implemented by using a simple rule based on the possibility distribution. The proposed method is confirmed through experiments.