Integrated Robotic System for Indoor Spatial Perception and Navigation for the Blind

Abstract

An interactive intelligent robotic system is presented to assist blind people with indoor activities. The integrated system employs practical human scene perception characteristics to achieve intelligent guidance through indoor environment. This system models the environment using topological maps and learns these cognitive maps gradually via a lead-through algorithm. A first-depth searching algorithm is implemented to achieve an optimized selected path. Multiple navigation sensors were utilized for obstacle detection and avoidance in addition to localization purposes. Sensors limitations and errors were corrected by implementing a sensor fusion algorithm using discrete Kalman filter. Also, visual landmarks were used as beacons for internal error correction in odometry and mapping. A comprehensive structural design of the robot is accomplished; besides, a human-robot interaction unit is designed using speech synthesis and recognition modules. The proposed system represents an integrated technology to assist the blind with daily indoor activities and to facilitate performing necessary personal tasks.