Fall detection for the elderly using a cane robot based on ZMP estimation

For assisting the elderly or physically challenged walking, a cane robot which be driven by an Omni-directional mobile robot has been proposed. A variety of sensors fixed on the cane robot is used for monitoring user's motion and detecting environment around user. The user's intentions of to which user want to move can be estimated by monitoring the applied force to cane robot, then the cane robot is controlled to move and aid user to walk. As a nursing-care robot, the safety is a most important concern; therefore a fall detection concept is proposed to estimate the risk of falls in elderly by referencing to the Zero Moment Point (ZMP) stabilization. An on-shoe sensor is proposed to estimate ZMP by obtaining dynamic reaction force at the contact of the foot with the ground. The safety of elderly's walking is defined in the case of that the ZMP is inside support polygon. If ZMP is on or out of the boundary of support polygon, the user will fall down or have a trend of falling down. Experimental results are presented success of proposed ZMP estimation and fall detection schemes.