Pedometer paper notes【DOC精选】.docVIP

Personal Navigation Using Novel Methods of Human Motion Modeling GPS provides a means of navigation and positioning, however it fails in many situations such and indoors and underwater. An understanding of how a person moves through an environment can provide contextual information that can aid in navigation in these GPS denied environments. This generic process depicted in Fig. 1 is a reasonable approach to more than just the motion classification as it can extend to any classification application. The process starts with a set of sensor data that is then framed and used to construct feature models. Next, a classification algorithm is applied, and finally, the results can be supplied to a navigation dead reckoning algorithm. There are a wide range of classification methods; binary trees, Bayes classifier, k-Nearest Neighbor, rule-based approaches, and Support Vector Machines (SVM) to name a few as used in [6]. Bayes classifiers assign a classification to the estimator with the maximum posterior probability and the K-Nearest Neighbor compares commonality between its k nearest neighbors. Artificial Neural Networks (ANNs) are also used in supervised approaches to classification. Classical implementations of ANNs have been applied to a wide range of IMU-based motion classification. More unconventional networks with dynamic topologies have been gaining in popularity. Hidden Markov Models and its variants have been used as a popular classification tool in speech recognition and have been adapted to number of other time series classification problems. For development on smartphones, processing load restrictions can limit some classification methods. Using a Three-Axis Accelerometer and GPS Module in a Smart Phone to Measure Walking Steps and Distance \ THREE-AXIS ACCELEROMETER SENSOR IN A SMART PHONE We can see that with both walking and running, their acceleration values will have a fixed and regular change, because when a human is walking, that basic