External Force Estimation for a Two-wheeled Mobile Robot Using a Deep Learning-calibrated Sensor System

Abstract

Mobile robots require compensatory driving measures for direct contact with people in blind spots that cannot be detected by sensors. In this study, we proposed an external force compensation interface device for robot applications to realize compensatory driving. The device can detect two-axis forces and displacements occurring on the application platform when placed between the driving platform of the mobile robot and the application platform assembled on the driving platform. The device was configured in the form of a thin plate placed between the driving and application platforms. We present a sensor calibration method to derive the optimal result based on the sensor arrangement method. An external force measurement module was developed using a deep learning method to determine the input angle and size of the external force. It was attached to a nonholonomic mobile platform to perform tasks for direct interaction. Calibration was performed using deep learning to evaluate sensor performance. We present a collaboration plan for the sensors and external forces of a nonholonomic mobile platform.