Efficient movement detection for human actions using triaxial accelerometer

Real-time sampling with high speed using a triaxial accelerometer is the fundamental purpose of using consumer devices over all-gesture operation. A welldesigned fall-down detection system uses a triaxial accelerometer influenced on gravitation to calculate an included angle between surfaces and axis of accelerometer and the accelerometer is spread on fixed position and pose of body. A triaxial accelerometer is difficult to employ to detect fall-down without postural orientation under an undetermined body position. This work presents a novel fall detection system based on unrestricted position, called Fall-Down Detection based on Unrestricted Position (FDUP), to provide an unrestricted measure for solving the undetermined position problem. The FDUP system provides an algorithm for unrestricted fall detection, which includes thresholds for acceleration and angle variation. Five volunteers were invited to simulate fall-down over 400 times to collect statistics, and to build the fall-down acceleration threshold. Analytical results show the feasibility of implementing a fall-down detection mechanism based on undetermined position.