—In this paper, we present the design and analysis of a portable position and orientation (pose) estimation aid for the visually impaired. Our prototype navigation aid consists of a foot-mounted pedometer and a white cane-mounted sensing package, which is comprised of a 3-axis gyroscope and a 2D laser scanner. We introduce a two-layered estimator that tracks the 3D orientation of the white cane in the first layer, and the 2D position of the person holding the cane in the second layer. Our algorithm employs a known building map to update the person’s position, and exploits perpendicularity in the building layout as a 3D structural compass. We analytically study the observability properties of the linearized dynamical system, and we provide sufficient observability conditions. We evaluate the real-world performance of our localization aid, and demonstrate its reliability for accurate, real-time human localization.
Joel A. Hesch, Stergios I. Roumeliotis