Advanced Map Matching Technologies and Techniques for Pedestrian/Wheelchair Navigation

Due to the constantly increasing technical advantages of mobile devices (such as smartphones), pedestrian/wheelchair navigation recently has achieved a high level of interest as one of smartphones’ potential mobile applications. While vehicle navigation systems have already reached a certain level of maturity, pedestrian/wheelchair navigation services are still in their infancy. By comparing vehicle navigation systems, a set of map matching requirements and challenges unique in pedestrian/wheelchair navigation is identified. To provide navigation assistance to pedestrians and wheelchair users, there is a need for the design and development of new map matching techniques. The main goal of this research is to investigate and develop advanced map matching technologies and techniques particular for pedestrian/wheelchair navigation services. As the first step in map matching, an adaptive candidate segment selection algorithm is developed to efficiently find candidate segments. Furthermore, to narrow down the search for the correct segment, advanced mathematical models are applied. GPS-based chain-code map matching, Hidden Markov Model (HMM) map matching, and fuzzy-logic map matching algorithms are developed to estimate real-time location of users in pedestrian/wheelchair navigation systems/services. Nevertheless, GPS signal is not always available in areas with high-rise buildings and even when there is a signal, the accuracy may not be high enough for localization of pedestrians and wheelchair users on sidewalks. To overcome these shortcomings of GPS, multi-sensor integrated map matching algorithms are investigated and developed in this research. These algorithms include a movement pattern recognition algorithm, using accelerometer and compass data, and a vision-based positioning algorithm to fill in signal gaps in GPS positioning. Experiments are conducted to evaluate the developed algorithms using real field test data (GPS coordinates and other sensors data). The experimental results show that the developed algorithms and the integrated sensors, i.e., a monocular visual odometry, a GPS, an accelerometer, and a compass, can provide high-quality and uninterrupted localization services in pedestrian/wheelchair navigation systems/services. The map matching techniques developed in this work can be applied to various pedestrian/wheelchair navigation applications, such as tracking senior citizens and children, or tourist service systems, and can be further utilized in building walking robots and automatic wheelchair navigation systems