A Hybrid Indoor Positioning System Using a Linear Weighted Policy Learner and Iterative PDR

Electronic indoor positioning systems deal with the combination of sensors, actuators, and computational algorithms for precisely locating subjects, delivering navigation directives, and keeping track of particular objects. The main factors considered for the construction and evaluation of these systems are the localization accuracy and the time spent to calculate and deliver this information. The challenge in developing successful positioning systems is to find a tolerable relationship between those factors. In this proposal, after a careful analyses of related works, we associated different methodologies and technologies to construct a hybrid positioning model that uses a mapping algorithm called Linear Weighted Policy Learner, a navigation model called iterative Pedestrian Dead Reckoning (which uses the Kalman filter to deliver real-time location), and an obstacle detection algorithm that combines sounds and stereo vision sensorial capabilities. The adopted choices were based on the published state-of-the-art, and comparisons of the obtained results showed that our system is accurate and fast enough to be very competitive with the current stage of the technology.