The survival of stroke often results in losing patient’s ability to perform limb actions needed for performing daily living activities. Most often, this is the result of damage to the part of the brain that is responsible for a certain action. Since it is a matter of plasticity, the patient cannot regain the same ability, but must be taught again. For the purpose of rehabilitation, rehabilitation robots can be used - most often in the form of an exoskeleton. The exoskeleton consists of inertial and EMG sensors that, according to the operation of the healthy hand (in the frame of this thesis), teach the robotic device the action that will be applied to the patient's hand. Inertial sensors are responsible for collecting data on linear acceleration, angular velocity and magnetic field strength. Precisely for this reason, they consist of a three-axis accelerometer, gyroscope and magnetometer. In the Precision Engineering Laboratory at the Faculty of Engineering University of Rijeka, commercial equipment and software manufactured by Shimmer Sensing company were used to describe the process of collecting and calibrating data obtained from IMU sensors. A few important conclusions were drawn during the work. The influence of noise on the received signal significantly affects the success of data calibration, which was shown in the data obtained from the accelerometer. Also, an important condition when measuring with an inertial sensor is that the measurement must be performed in a space with no local magnetic fields that could affect the signal. Finally, additional data filtering is suggested for successful processing and use of the same.