In the course of writing my thesis I joined the developing of a balancing robot developed on the MIT department of BME-VIK. At the beginning of my work the robot was already in working state, the control algorithm could roughly hold it in balance, but we wanted to improve the algorithm to be more effective and able to get the robot back in balance even if it gets a larger disturbance or a quick speed-change. As the base of the controlling algorithm is the accurate knowledge of the angle of the robot according to the vertical plane – without this the controlling algorithm can not work properly – I searched for how to make the estimation of the angle more accurate. In this thesis I review the possible solutions for angle estimation with different equipments and methods, the difficulity and disadvanteges of these methods, I describe the methods implemented so far and the cases they have neglected. Finally I try to find a more accurate algorithm for angle estimation using some dinamic models of the robot.