Parametric Design and Analysis of the Arc Motion of a User-Interactive Rollator Handlebar with Hall Sensors

Most user-interactive rollator handles fabricated so far have had errors in classifying users’ walking intentions due to the structure not considering gravity, in addition to their usage not being intuitive. Here, we introduce a smart handle based on ‘arc motion’ horizontal to the ground to classify the user’s walking intentions accurately. In designing the handle, the limit of grip angle is adopted considering the arc motion. This minimizes the deflection of the handle by gravity and can allow the handle to be optimized for the grip motion. Moreover, we applied the results of our analysis of users’ walking intentions in the arc motion to machine learning. In understanding the user’s intentions to walk, we created two support vector machine classifiers from handle data collected through four Hall sensors. This combination of the arc motion and machine learning has significantly reduced classification errors. As a result, we ensured an accuracy of 0.95, which is a widely used standard in control.