This study presents an approach to improving the localization of mobile robots in a controlled test environment through sensor fusion. Currently, the position of the ground robot (Unmanned Ground Vehicle (UGV) P3DX) can be estimated using an optimization algorithm or another technique based on a previously known map. In contrast, the aerial robot (Unmanned Aerial Vehicle (UAV) Bebop Parrot 2) can determine its relative position by detecting a marker on the ground robot using the Robot Operating System (ROS) WhyCon package. However, ambiguities or disturbances may compromise the accuracy of the robots' localization. To mitigate these limitations, Ultra-WideBand (UWB) sensors were incorporated into the system, enabling a more robust data fusion by integrating information from multiple sensors. Additionally, a filter was applied to reduce the impact of high-frequency noise on position estimation. Accurate localization test data were used to construct a simulated scenario and analyze the performance of the proposed approach. The simulation results demonstrate that sensor fusion, combined with noise filtering, significantly improves localization accuracy, making the approach promising for applications in mobile robot navigation.