Dynamic localization based on EKF for indoor mobile robots using discontinuous ultrasonic distance measurements

Abstract

We present an efficient dynamic localization based on EKF (Extended Kalman Filter) for indoor autonomous mobile robots using multiple ultrasonic distance measurements, which is a fundamental technique to perform robots intended navigation tasks without human intervention. We present an EKF-based algorithm with a state/observation vector composed of the robot pose using odometric and ultrasonic distance measurements. Our ultrasonic localization system consists of several ultrasonic Txs fixed at ceiling (with known positions in global coordinates) and three Rxs on the top of the robot (with 2-D isotropic ultrasonic Rx array). We also propose a dynamic distance estimation method, which can compensate the ultrasonic distances from the odometric information traveled by the robot, from at the instant to obtain actual ultrasonic measurements to at the instant the localization to be performed. Our hybrid self-localization algorithm utilizes these distance estimation data, which enables accurate determination of the robots pose irrespective of its motion. Experimental results show that the proposed method is more accurate than the conventional method.