Guidance System for Visually Impaired Persons Using Multi-Sensor Fusion

Abstract

We propose a guidance system using a sensor fusion technique, to safely guide visually impaired person to the destination. To assist users in walking, it is important to accurately and quickly estimate their current position. We introduce methods to estimate position of the user in buildings by using an infrared camera and markers, and to estimate their location outdoors by exploiting rangebased ZigBee and GPS (global positioning system) data. At the same time, we calculate relative position against the specific location and orientation of the user by employing accelerometer and gyro sensor data. In addition, we developed an algorithm that analyzes range data obtained from multiple ultrasonic sensors to avoid obstacles on the user's path. The genetic algorithm is used to design the sensor network with optimal placement of ZigBee modules. We made a walking guidance system for the visually impaired with these methods.