Multiple Sensor Fusion in Mobile Robot Localization

Abstract

The fusion of multi-sensory information plays a key role in driving a mobile robot over a fixed lane, object recognition, obstacle avoidance, self localization and path planning. To learn the environment using multi-sensory information, we need both an accurate sensor model and a reasonable sensor fusion methodology. In this paper a novel technique is explained combining data's from ultrasonic sensor, encoder and gyroscope. Encoder is often utilized for the position estimation by accumulating the number of times the wheel rotates. Since the B21r robot relies only on encoder data information for localization, the motivation behind this technique is to reduce the wheel-drift that occurs in encoder due to slippage error and bumps on the path that causes the robot to move in a elliptical path when intended to move in a circular path. The B21r mobile robot has forty eight ultrasonic sensors, twenty four at the base and twenty four at the body of the robot. The ultrasonic sensors are used to develop an obstacle avoidance algorithm based on virtual force field (VFF) technique. The algorithm is then combined with a rule based algorithm for the inertial sensors namely encoder and gyroscope, which switches the control back and forth between the encoder and the gyroscope depending on the slippage error caused in the encoder. The simulation results show the path of the robot with the conventional encoder data alone and then with the algorithm implemented. The results and future expansion of the study and the merits of the algorithm are discussed.