Analysis of Systematic and Random Coordinate Errors of Visual Landmarks and Their Effect on Positioning Algorithm Accuracy

Visual landmark-based positioning systems are becoming increasingly popular in mobile robotics, autonomous vehicles, and indoor navigation technologies. One of the key factors determining their accuracy is the correctness of landmark coordinates, which in practice can be distorted by both systematic offsets and random noise. This requires a quantitative assessment of the impact of such errors on the operation of positioning algorithms. Subject of research: analysis of the impact of systematic and random errors in determining the coordinates of visual landmarks on the accuracy of positioning algorithms. The research addresses the assessment of sensitivity in different positioning algorithms to offsets and noise in landmark data and to identify critical factors that most affect localization accuracy. Methods applied: simulation modeling with the ability to vary the parameters of systematic and random errors, reproduction of four scenarios (no errors, only bias, only noise, combination). The mean absolute error (MAE) and root mean square deviation (RMSE) were used to assess accuracy. The following results were obtained. Even small errors in the coordinates of landmarks significantly reduce the accuracy of positioning. It was found that systematic errors have a more critical impact on the results compared to random noise. The centroid and weighted centroid methods were the most resistant to errors, while lateration showed high sensitivity to systematic shifts.