Flexible Capacitive Touchpad for Real-time Gesture Recognition in Automotive Lighting

Abstract

Abstract. In the automotive industry, the shift towards controlling various car functions via a touch screen on the dashboard offers many contemporary experiences. In particular, integrating multiple control operations on a single component, such as buttons and touch screens, is increasingly popular. The combination reduces the complexity associated with mechanical buttons on the vehicles while still ensuring full functionalities. In this paper, we propose a system that uses flexible sensors with conductive threads attached directly to the car's leather interior. The goal is to collect the necessary information through capacitance changes during various touch operations to control the automotive system. Besides, the touch signal results are enhanced using a ID convolutional neural network (ID-CNN) algorithm. The ID-CNN model can recognize 15 types of touch actions with an accuracy of 99.47% and an average recognition time of 2.025ms. More specifically, the small-size CNN model can be applied on embedded boards with limited hardware resources. This incredible capability offers excellent potential in the field of the Internet of Things (IoT) or embedded machine learning.