Research On Zero-Velocity Update Technology Of Pedestrian Inertial Navigation Based On MEMS

With the development of technology and the rapid advance of urbanization, massed high buildings and large enclosed indoor environments becomes more and more. As signals will be weakened severely when they go across buildings, it’s difficult to apply traditional GPS positioning technique into indoor navigation and positioning. However, the rapid development of MEMS brings a solution to this problem. For its small size, low cost and low power consumption, inertial device based on MEMS has been studied extensively in the area of indoor inertial navigation applications. But, as the MEMS inertial device has inherent drift, noise and other errors, how to eliminate the errors is an important problem when it is used in indoor navigation.This article intends to do a research which is based on foot-mounted inertial navigation system on zero-velocity update for its own noise error and the drift error introduced by movement under different pedestrian gaits (normal walking, running). The principle is that when pedestrian moves in different gaits, there will exist a period of time that the speed is zero in theoretical in the process of foot contacting with the ground, but the actual value of the speed is not zero in fact. So, the error value can be predicted by detecting the zero speed moment at the pace time, and then it can be removed. The detection of the zero-velocity moment will be based on the inertial sensor data (acceleration and angular velocity values) and the distance between ultrasonic module fixed at the top of instep and ground respectively. For the zero-velocity detection algorithm on the basis of inertial sensor data, it adopts a multi-condition approach based on acceleration squared, variance of acceleration squared and angular velocity. While for the zero speed detection algorithm according to ultrasonic data, it is to deduce the ultrasound data model first on the analysis of the feet’s gait model, then summarize the detection algorithm according to the model, thus detect the zero-velocity moment. Considering the stability of the ultrasonic data and in order to detect the zero-velocity moment easily, locally weighted scatterplot smoothing algorithm is used to process ultrasound data, which makes it to show the patterns and the trends of motion model more specifically on the whole. After detecting the zero-velocity moment by both kind of zero-velocity detection algorithm, the Kalman Filter will be triggered to predict errors and update both the position and attitude information.The feasibility of zero-velocity detection algorithm is demonstrated by experiment. The results show that at different gait, the steps, single step length and the whole distance obtained by zero-velocity update based on inertial data and ultrasound data match practical case. Both zero-velocity detection methods can detect the number of steps100%, and also there is little difference between single-step length and the actual value. After correcting, the whole distance error is below1%in normal walking gait while it is below2%in running gait. But the result of zero-velocity update based on ultrasonic data is better than that based on inertial data.In this paper, different aspects of analysis and verification are made for the zero-velocity detection algorithm of the zero-velocity update technique, which provides a simple and efficient method for the application of MEMS inertial sensor in indoor navigation.