On improving low-cost IMU performance for online trajectory estimation

We have developed an automatic mitigation method for compensating drifts occurring in low-cost Inertial Measurement Units (IMU), using MEMS (Microelectromechanical systems) accelerometers and gyros, and applied the method for online trajectory estimation of a moving robot arm. The method is based on an automatic detection of system’s states which triggers an online (i.e. automatic) recalibration of the sensors parameters. Stationary tests have proven an absolute reduction of drift, mainly due to random walk noise at ambient conditions, up to ~50% by using the recalibrated sensor parameters instead of using the nominal parameters obtained from sensor’s datasheet. The proposed calibration methodology works online without needing manual interventions and adaptively compensates drifts under different working conditions. Notably, the proposed method requires neither any information from an aiding sensor nor a priori knowledge about system’s model and/or constraints. It is experimentally shown in this paper that the method improves online trajectory estimations of the robot using a low-cost IMU consisting of MEMS-based accelerometer and gyroscope. Applications of the proposed method cover automotive, machinery and robotics industries.