Multi-People Pose Identification System

At my first video, my research work during my graduate study in 2007 was to develop a module of detection of human poses. The module is with the user waist belt and capable of finding seven types of human poses, e.g. standing, sitting, lying, walking, falling off, and going upstairs and downstairs. The module is composed of a three-axis-accelerometer, a Micro-Control Unit (MCU), and a Zigbee communication chip.

The detection procedure of the module is as follows. First, MCU collects data from the three axial accelerations through A/D converters within 2 seconds. Second, MCU processes an algorithm to analyze the data about signals and results in a possible human pose. Third, MCU transmits the possible human pose to a receiver through the Zigbee chip, and then the first step is returned and so forth. At the receiver module, the possible human pose sent by MCU is obtained via a Zigbee interface. The receiver module transmits it to a computer through RS-232. I also wrote a program in the computer to show the result and do some statistical work. My developed system can also be used to detect multiple people poses with a potential application for the elderly healthcare. For example, if the elderly carelessly falls off, his/her poses can be detected, and then nurses or doctors are remotely alerted to give assistance properly. This developed system has been patented in the US and Taiwan.