Modelling human falling process based on a five-link bipedal robot

In this paper we develop a dynamic model to mimic and simulate human falling process when tripping over an obstacle. The work consists of three parts. First a 5-link bipedal model is established, the joint trajectories of knee and hip are generated. Next several constraints are taken into consideration, which take the physical, physiological and environmental constraints into consideration. Hence the fall process can be modeled in a realistic manner. A number of important factors that characterize the fall process, such as reaction time, inclining angle, stable range, initial speed, are investigated. The whole falling process is formulated according to biological characteristics and stability analysis, which further offers an effective way to identify and distinguish falling from normal activities, so that the falling can be predicted and a prevention device can be activated in advance.