Evaluating possible errors in IMU sensors

Sensors with inertial working principle are used in many different fields, especially in human modeling and navigation applications. Accelerometer and gyroscope sensors, which are the main components of inertial sensors, can be produced with micromechanical system technology today. A decrease is observed in their performance as well as low cost and high production amount facts. Therefore, various error compensation alogirthms have been proposed to improve inertial sensor performances. Deterministic and probabilistic inertial sensor errors and their relationships with the parameters that cause these errors from the basis of error compensation algorithms. In order for error compensation algorithms to perform high accuracy error compensation, it is critical to establish error modeling that reveals the relationship between sensor errors and error sources. In this thesis, a test device capable of simulating human arm and leg movements was produced in order to model inertial sensor errors. Three different axes of inertial sensors at two different angular velocities were applied separately with a human motion simulation test device, and sensor error distributions were calculated using the Gaussian distribution model. The effect of angular motion physical effects and stability of sensor temperature on inertial sensor error distributions was investigated using the test device.