Proposal of a method for estimating gripped object mass in stoop lifting using a 9-axis sensor

In recent years, low back pain(LBP) among manufacturers has become a problem. One causes of LBP is the lifting of loads. It has been confirmed that the stoop lifting, where the object is lifted by bending forward with extended knees, puts more strain on the lower back than squat lifting, where the back is straight and the knees and hips are flexed. In addition, it is recommended that men over 18 handle loads of about 40% or less of their body weight, so knowing the object’s weight is important. In this study, we propose a method to determine if the weight of the grasped object is 10 kg or 20 kg immediately after lifting, using a 9-axis sensor attached to the chest and the stoop lifting, which is considered to have a large load on the lower back. A linear approximation was made based on the most recent 0.2s data of the forward tilting angle calculated from the 9-axis sensor measurements, and the change in the tilt was obtained by repeating this process. The difference and maximum of the extreme values at the start of lifting were calculated from the change in inclination, and the amount of grasped material was discriminated against by k-fold cross-validation using a nonlinear Support Vector Machine (SVM) with these two as the feature values. The discrimination accuracy was 82.3%, 7% higher than that of the conventional method calculated under the same conditions. It is necessary to conduct further validation using different lifting methods, known masses, and smaller mass differences in the future.