A personal navigator is typically described as a system designed to provide continuous navigation of a mobile user within a certain positioning accuracy. Personal navigators have been studied for about a decade in different fields and applications, such as visual surveillance, rescue operations, security, emergency, police safety, and military applications.

The common goal of all these applications is to provide precise and reliable position/velocity/heading information of the individuals in various environments. In the open sky environment, either GPS alone, or a GPS/IMU (Inertial Measurement Unit) system can facilitate the basic navigation functionality with the accuracy depending on the choice of GPS and IMU sensors. In confined environments, however, the main challenge for a personal navigator is to implement a backup plan to maintain the navigation information in the absence of GPS signals.

A variety of alternative methods, based on different sensors with different potential applications, can facilitate positioning in real time in areas where GPS signals are blocked. These systems offer either absolute or relative positioning capabilities. One example is a synthetic vision system that can be used to reconstruct the motion trajectory. The basic tracking idea is to match the captured images with the pre-registered images stored in a database, but the online computational burden of such a system is very heavy with respect to available computer power. Image-based navigation can also use a sequence of images collected by a mobile user (with no reference imagery) to reconstruct the navigation trajectory after a loss of GPS. In that case, the last image with GPS-based georegistration provides a connection to the subsequent image sequence that allows only for relative navigation solution. By connecting to the last known GPS-based position an absolute navigation can be achieved. Instead of a vision-guided system, in a wireless and smart environment one can use wearable sensors and employ techniques such as infrared-based beacon systems, Radio-Frequency Identification RFID-tags, Wireless Local Area Network (WLAN) and ultra wideband (UWB) systems, or cellular phone positioning for absolute position determination. Pseudolite technique and the high-sensitivity GPS receivers also can be effectively used for personal navigation.

However, in many applications, such as emergency response and military operations, it is often impossible to prepare the environment in advance to fit the need of a personal navigator different than a GPS-based one. To this end, in uncontrolled environments, a navigation system should rely only on self-contained sensors, such as accelerometers, gyroscopes, digital barometers, electromagnetic compasses, and step-sensors to deliver relevant parameters required for dead-reckoning (DR) navigation: direction, walking distance, and altitude.

Human Body as Nav Sensor

Human motion is a complex process affected by a large number of factors, including body characteristics, a person's physical condition, his or her locomotion type, and terrain and environmental conditions. Despite these dependencies and a variety of motion dynamics, there are repetitive patterns in sustained motion. Thus, exploiting human locomotion as a sensor offers additional information to a personal navigation system.

Developments in the past few years in machine learning techniques have led to an exponential increase in engineering, social, and biomedical applications that make use of machine learning. In particular, machine learning techniques are suited for applications related to human motion modeling and are being increasingly used for this purpose, due mainly to the complexity of the biological systems as well as the limitations of the existing quantitative techniques in modeling.

Using machine learning methods allows for better process control and more reliable prediction and modeling of the processes under consideration. Examples of algorithms and methods used in machine learning are Artificial Neural Networks (ANNs) and Fuzzy Logic (FL). We use the ANN and FL methods to create a simplified human dynamics model that consists of three basic parameters: step length (SL), step frequency (SF), and step direction (SD). Together, these parameters are used to navigate the mobile operator in the dead-reckoning mode. The human dynamics model is calibrated when other sensors, primarily GPS, provide a continuous navigation solution, and the human-based navigation is activated when GPS is significantly degraded or not available.