Analysis of Odometric Localization Uncertainty Using a Motion Capture System and Monte Carlo Simulation

In recent years, there has been a very dynamic development of technologies related to mobile robots, especially autonomous robots. One of the basic competences of mobile robots is navigation and related localization. The article presents the results of research aimed at estimating the accuracy of localization using the differential drive kinematics model. Experimental and simulation studies of the QBot 2e mobile robot were performed. As part of the experimental studies, hardware-in-the-loop simulations of the robot’s movement along the given paths were carried out with simultaneous speed measurements. The measurement results were used to conduct Monte Carlo simulations aimed at estimating the uncertainty of determining the robot’s position. Based on the research results, it was found that the uncertainty of localization increases with time and the probability densities of random variables representing the robot’s position can be approximated with high accuracy by the Gaussian distribution.