Updating Position Estimate of Mobile Robots in Curvilinear Environments via Range Measures

This paper addresses the problem of updating a rough, odometric estimate of the pose of mobile robot, by means of a single range measure. The linearization errors, typical in curvilinear environments, often make the Extended Kalman (EK) estimate unreliable. Therefore, a second-order method is proposed, in which the measurement result is approximated in terms of a two-degree function of the pose estimate error (instead of a linear one). Experiments conducted on a real robot navigating within a curvilinear environment are reported, showing that the errors in the updated estimate, provided by the proposed method, are signiicantly lower than the errors in the EK estimate.