Fall Detection System on Smart Walker Based on Multisensor Data Fusion and SPRT Method

Traditional walkers are commonly used for the elderly in social life, which solves basic problem of walking, but it is difficult to ensure safety when a fall occurs, and human-computer interaction poor. The image recognition method or IMU sensor fixed on user, such as wearable watch, by most current detection methods. Wearable sensors require user’s wearing operation, little troublesome, accuracy related way wearing. requires high-priced camera installation position, unable adapt outdoor activities. We investigated low-cost mounting body walker. propose this paper an improved method, namely Precondition Limit Threshold SPRT (PLT-SPRT), novel system smart walker based PLT-SPRT. signals upper lower limb fused Kalman filter algorithm, admittance control parameters obtained through identification method. In study, sequential probability ratio test algorithm set null hypothesis alternative hypothesis, construct likelihood optimize decision function, judge whether falls occur. simulated Matlab software, user intention after fusion more accurate, optimized function judged accurately. Verified embedded STM32 equipment real world, can accurately identify fallen state, with low delay, state detected about 160ms earlier than traditional threshold-based at same time, higher 94.9%, meets high real-time requirements ideal solution walkers.