Akademik Veri Yönetim Sistemi
Acta Polytechnica Hungarica, cilt.23, sa.2, ss.7-24, 2026 (SCI-Expanded, Scopus)
Localization is one of the most critical technical problems of autonomous robots. To evaluate the performance of any localization algorithm, the ground truth data is required for comparison. On the other hand, high-cost, complex, and limited-area effective, precise camera systems are utilized to obtain accurate position information. In this study, a novel device has been developed and produced as an alternative to existing systems. It is simple, low-cost, easily integrated into many robots, and capable of leaving a mark directly on the ground, making it effective throughout the robot's motion regions. During the testing process, the position of each mark was measured manually to confirm the accuracy of the measurements taken by the device, which was mounted on the autonomous wheelchair and automatically left marks on the ground, at certain numbers and intervals. The data obtained with the device have been compared with Adaptive Monte Carlo Localization (AMCL), a localization algorithm, and Odometry data. As a result of this comparison, it has been observed that the Odometry-based prediction vector initially closely matched the ground truth data but accumulated errors over time. In the AMCL algorithm, however, although instant position errors may occur using Lidar measurement data taken from the surroundings, these errors have been determined to decrease significantly in the next steps. As a result, it has been confirmed that the obtained results are compatible with theoretical expectations.