Inertial Systems

Abstract

The sensors available for tracking systems can be loosely broken into groups based on the measurement reference frame: (1) idiothetic or body reference sensors such as inertial sensors and encoders which measure motion of the body in the body frame; and (2) allothetic sensors or external reference sensors such as magnetic field, processed GPS and pressure which measure heading, 2D location and ele-vation in the global or Earth frame, and local reference sensors which measure ranging or bearing to local reference points/landmarks (Chaps. 2, 9). In this chapter, we focus on the use of inertial idiothetic sensors as part of a pedestrian location and tracking system in GPS denied areas. Inertial measure-ments are differential measurements in the sense that they quantify changes in speed or direction. The two primary types of inertial sensors are accelerometers and gyroscopes. Accelerometers measure instantaneous changes in speed, or equivalently force, and gyroscopes provide a fixed frame of reference with which to measure orientation or equivalently change in direction. As such, any navigation information obtained from the sensor system is used to compute movement rela-tive to the starting location and orientation. A navigation system that estimates current position from a prior known position and measurements of motion (for example, speed and heading) and elapsed time is called a dead reckoning system. A serious disadvantage of relying on dead reckoning for navigation is that the errors of the process are cumulative without other externally referenced correc-tions such as GPS, compass, or landmark-based corrections. Nevertheless, dead reckoning techniques have been used for decades by the Department of Defense, NASA, and others for sophisticated navigation systems (Titterton and Weston 2004). These systems have relied on very high quality mechanical, fiber optic, or laser sensors with size, weight, power, and cost beyond the range of what is needed for consumer applications. Table 8.1 gives an idea of the error growth performance requirements for different classes of inertial measurement units (IMUs). The limits on error growth set requirements on the types of sensors that can be used. The ability to fabricate microelectromechanical systems (MEMS) using semi-conductor device fabrication technologies (Ghodssi and Lin 2011) has lead to the C. Gentile et al., Geolocation Techniques, DOI: 10.1007/978-1-4614-1836-8_8, Ó Springer Science+Business Media New York 2013 213 development of low size, weight, power, and cost MEMS inertial sensors. Using these consumer grade navigation sensors that have only recently become an option in cell phones, new data is available that can be leveraged to improve location accuracy for on foot personnel in GPS denied or degraded areas. Already cell phone applications include the ability to enhance location using cell carrier location services (cell tower triangulation—Chap. 5) and Wi-Fi (provided by Skyhook and now Apple, Google, and others—Chap. 4). The motivation for this chapter is to review how the new consumer grade sensors can be used to improve pedestrian tracking. The chapter is organized as follows: First, we discuss some of the limitations of GPS as a sensor for pedestrian tracking. Then, we discuss MEMS inertial sensors and review some of the standard computations that are used with general inertial navigation sensors. Pedestrian tracking is unique relative to vehicle tracking where dynamic models are available and control inputs are known, so next we review specific approaches to imple-mentation of pedestrian navigation systems. Heading errors are the largest source of error in inertial pedestrian navigation systems so we divert the discussion from inertial sensors alone to include a dis-cussion of the use of magnetic sensors for heading correction. It is important to note that using magnetic sensors for heading correction indoors is not straight forward because of the common occurrence of large magnetic disturbances caused by electrical systems and the building structure itself. We conclude the chapter with a discussion of accuracy metrics. It is typical to see inertial navigation system error quoted as a percent of distance travelled; however, this can be very misleading as it does not account for heading errors which are typically the largest source of error.