Modeling and analysis of wearable low-cost MEMS inertial measurement units

This paper presents a minimized wearable inertial measurement units (IMU) which consists of a three-axis accelerometer, a three-axis gyroscope, a three-axis magnetometer. These sensors are Micro-electromechanical Systems (MEMS) devices, which have advantages of lower cost, smaller size and lower power consumption. The size of MEMS IMU is 23×23×4mm that is small enough to be easily fixed on hand, waist or shoe for monitoring patients' behavior, tracking people's indoor trajectory and playing games. However, low-cost MEMS inertial sensors are characterized by high noise and large uncertainties in their outputs such as bias, scale factor and errors. We implement the Allan variance method to determine the characteristics of the underlying of various types of error terms of random processes. The MTi-300 is a commercial industrial grade MEMS IMU, which has high precision and low noise. We collected data from wearable IMU and MTi-300 staying on turntable in static state for one hour, then we transfer the data to a laptop and analyze them in MATLAB software. We give comparison analysis about their Allan variance results. In future, the wearable MEMS IMU can get attitude determination in gaming, industrial and medical with the less error and low noise model.