Localization of a Mobile Autonomous Robot Using Extended Kalman Filter

This paper demonstrates an effective method for combining measurements from a gyroscope and rotary wheel encoders (odometry) in mobile robot localization. Sensor fusion of this kind is done using an Extended Kalman filter obtained from the values of above sensors for a mobile autonomous robot. Many such methods implement a statistical model that describes the behaviour of the gyroscope and the odometry component. However, because these systems are based on models, they cannot anticipate the unpredictable and potentially "catastrophic" effects of irregularities and frictional changes occasionally encountered on the floor. We present experimental evidence that non-systematic odometry error sources impact the robot's motion. Therefore a new approach has been developed based on a study of the physical interaction between ground and the robot. This approach has been implemented by developing an embedded system with ARM 7 based LPC2148 micro-controller. Experimental results show that the proposed method effectively reduces the localization error while yielding feasible parameter estimation.