Dead-Reckoning Method for Personal Navigation Systems Using Kalman Filtering Techniques to Augment Inertial/Magnetic Sensing

Abstract

Sensing techniques and computer algorithms help relieve demands in answering three questions that are crucial to navigation problem solving: “where am I?”, “where am I going?”, and “how should I get there?”. The question “where am I?”, namely localization, is addressed in this paper, in regard to the development of personal navigation systems (PNS) for determining a human walker’s position and orientation. Biomedical applications of PNS technology, the ones we are mainly interested in, may include ambulatory monitoring systems for assessing the physical activity behavior of persons with disabilities and chronic health conditions, orientation & mobility aids for the blind or visually impaired, assistance services for frail seniors with compound problems of memory loss and disorientation. In principle, PNS would provide the localization data in any environment, and at any time. Design criteria, in terms of, e.g., portability, accuracy, availability, cost, hindrance to the natural walking pattern, and the idiosyncrasies of the environment where the user walks greatly concur in practice to restrict the choice of sensing techniques and computational methods suited to implement PNS. Localization in outdoor environments is easy to solve using Global Positioning System (GPS) technology, despite that a number of serious shortcomings exist: loss of satellite track, when direct line of sight with the GPS constellation is precluded by obstructions, inability to provide static heading information, significant power consumption. When localization in indoor environments is pursued and GPS is therefore useless, other externally-referenced sensing techniques are available (video movement-sensing, infrared, ultrasound); however, their operation is typically based on complex and costly measuring hardware, dense environment infrastructure, not to mention the severely limited properties of the external sources themselves, in terms of, e.g., range, field-of-view and so forth. Internally-referenced sensing techniques, namely inertial sensing, used in association with a relative-measurement approach, namely dead-reckoning (DR), can be a useful navigation alternative for implementing self-contained PNS (Fang et al., 2005). Being internally referenced and immune to interference and shadowing, inertial sensors (accelerometers and gyros) sense movement, in principle, without restrictions in the spatial domain (Welch &