High Precision Localization of Mobile Robots in Indoor Environments With UWB and Odometry
Makaleler > Tam Makale · 2026
Yazarlar: Mina Khoshrangbaf, Murat Aydemir, Nusin Akram, Gül BOZTOK ALGIN, Geylani KARDAŞ, Vahıd AKRAM
IEEE Access, Uluslararası, Wos, 2026, cilt 14, ss 87496-87507
DOI: 10.1109/access.2026.3701954
Özet
Determining the real-time and accurate location of mobile robots within indoor environments is crucial for a wide range of applications. Due to the limitations of GPS signals, alternative localization techniques are needed to determine the real-time location of mobile robots in indoor environments. This paper investigates the accuracy of odometry and Ultra-Wideband (UWB)-based localization methods for detecting the location of mobile robots navigating complex trajectories. Odometry relies on wheel rotations or movement sensors to estimate a robot’s position relative to its starting point, while UWB technology utilizes short-range radio waves to achieve precise distance measurements between UWB-equipped devices. We propose an adaptive localization method that combines the coordinates reported by odometry and UWB systems using a Kalman filter-based approach. The method dynamically adjusts the contribution of each system based on their estimated error rates, relying more on odometry when UWB experiences temporary interference, and shifting weight toward UWB as odometry drift accumulates over time. To evaluate the accuracy of the proposed method, we conducted various experiments in a testbed environment with six UWB nodes and mobile robots equipped with odometry sensors. Various trajectories, including circles, squares, triangles, and complex shapes, were tested. To evaluate the localization accuracy, combined coordinates were generated by integrating different proportions of odometry and UWB outputs. The results indicate that the proposed method determines the location of the mobile robot with between 13.06% and 33.59% lower average error lower average error compared to UWB, odometry, and their different combinations.