This document presents the design and implementation of two state estimators for a segway-like balancing robot using Kalman filtering techniques: the Extended Kalman Filter (EKF) and the Unscented Kalman Filter (UKF). The study analyzes their performance on various datasets, highlighting the impact of noise in state and measurements on estimating the robot's states, particularly the angle and angular velocity. Results indicate that both EKF and UKF effectively track the robot's state, though bias estimates exhibit some discontinuities.