Machine Learning for Capacitive Proximity Sensor Data

In order to increase safety in human-robot interaction, it is important to find reliable methods and approaches allowing better detection of people and objects in the robot environment. One research area concerns the development of proximity sensing technology. This paper intends to address this challenging requirement by applying a machine learning model specifically dedicated to a new kind of capacitive tactile proximity sensor (CTPS) that has been developed by the Intelligent Process Automation and Robotics Lab (IPR-KIT) in Germany. Our research study focused on using machine learning approach to infer from the data collected by the sensor array in order to insure a safer human–robot interaction. To achieve our goal, we built a classifier and a regressor on the projected distance between objects and the sensor.