3D global acceleration estimation using inertial measurement unit for biomechanical research
Konda, Revanth Raghuram
MetadataShow full item record
The effects of mechanical vibrations on the body of a driver in off-road vehicles such as mining vehicles have become one of the intriguing research topics. To reduce the injuries of the operator of an off-road vehicle where vibrations of high magnitude occur, an in-depth study of the nature of these vibrations must be done. The first step towards this objective is the correct estimation of acceleration of the vehicle by taking into consideration all the translational and rotational motion. The main goal of this study is the estimation of acceleration of an off-road vehicle in 3D space using an inertial measurement unit (IMU). In the past biomechanical studies, IMU along with a digital low-pass filter such as Butterworth filter has been used to estimate acceleration, but the estimation was only limited to vertical Z-direction. Also, the use of a low-pass filter requires deciding a threshold frequency by trial and error. In this study, to obtain the best estimate of three-dimensional acceleration, an algorithm is proposed using 3D transformations, Fourier analysis, and magnitude-based filtering methods. The use of magnitude-based filtering eliminates the trial and error process of selecting the threshold frequency for filtering and filters out the white noise and other sources of error present in the measurement signal. A preliminary experiment was first conducted in order to check the accuracy of the IMU used in this work. In the experiment, one-dimensional vibrations were produced in the vertical Z-direction using a heavy-duty excitor and the position was estimated using the IMU. The generated vibrations were sinusoidal with a frequency of 3Hz and a peak acceleration of 2.3 m/s². The estimated data using the IMU had a peak displacement of 2.38 m/s² with 3Hz frequency. A final experiment was conducted to verify the proposed algorithm in which a three dimensional motion including rotation about Z-axis was generated at three dierent frequencies using a 6-DOF robot (Adept 850s Viper). Based on a detailed error analysis, the final experiment results indicate that the proposed algorithm produces an acceptable degree of accuracy in estimation of 3D global accelerations.