Robust location using system dynamics and motion constraints

Abstract

To our knowledge, the indoor location system which currently achieves the best performance using inexpensive off-the-shelf equipment locates a mobile within 1.5 meters with probability 77% in hallways. Even while maintaining this accuracy, the system often reports logical errors such as the mobile in the wrong cubicle of an office or even on the wrong side of a wall when broadening the domain of application to within rooms. We propose an extension of the work using the same Markov localization framework, however incorporating system dynamics (necessitating no post-processing of the output) and motion constraints which implicitly encode the physical properties of the survey area. Our system retains the advantages of its predecessor of low cost, wireless LAN connectivity and security, and large-scale deployment, however extending the survey area from simple hallways to the whole office environment, while maintaining the same precision without logical errors.