EFFICIENT INDOOR LOCALIZATION VIA REINFORCEMENT LEARNING

In recent years the widespread use of Global Positioning Systems (GPS) has enhanced our abilities to find our way in outdoor environments, dispensing our over-reliance on visual features. However, GPS technologies do not solve the problem of localization and way finding in indoor environments due to transient phenomena such obstacles and fading. With the development of computer vision and deep learning techniques, vision-based localization and way finding have recently received special attention in the scientific community. In this paper we propose a method which learns to localize with high accuracy while minimizing the total number of processes based on reinforcement learning. We train our model on a dataset at MIT campus and we evaluate its performance by comparing with state-of-the-art techniques, obtaining higher accurate results.